MOTIVATION: Synthetic lethality (SL) is a promising form of gene interaction for cancer therapy, as it is able to identify specific genes to target at cancer cells without disrupting normal cells. As high-throughput wet-lab settings are often costly and face various challenges, computational approaches have become a practical complement. In particular, predicting SLs can be formulated as a link prediction task on a graph of interacting genes. Although matrix factorization techniques have been widely adopted in link prediction, they focus on mapping genes to latent representations in isolation, without aggregating information from neighboring genes. Graph convolutional networks (GCN) can capture such neighborhood dependency in a graph. However, it is still challenging to apply GCN for SL prediction as SL interactions are extremely sparse, which is more likely to cause overfitting. RESULTS: In this article, we propose a novel dual-dropout GCN (DDGCN) for learning more robust gene representations for SL prediction. We employ both coarse-grained node dropout and fine-grained edge dropout to address the issue that standard dropout in vanilla GCN is often inadequate in reducing overfitting on sparse graphs. In particular, coarse-grained node dropout can efficiently and systematically enforce dropout at the node (gene) level, while fine-grained edge dropout can further fine-tune the dropout at the interaction (edge) level. We further present a theoretical framework to justify our model architecture. Finally, we conduct extensive experiments on human SL datasets and the results demonstrate the superior performance of our model in comparison with state-of-the-art methods. AVAILABILITY AND IMPLEMENTATION: DDGCN is implemented in Python 3.7, open-source and freely available at https://github.com/CXX1113/Dual-DropoutGCN. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SUMMARY: The molecular changes induced by perturbations such as drugs and ligands are highly informative of the intracellular wiring. Our capacity to generate large datasets is increasing steadily. A useful way to extract mechanistic insight from the data is by integrating them with a prior knowledge network of signalling to obtain dynamic models. CellNOpt is a collection of Bioconductor R packages for building logic models from perturbation data and prior knowledge of signalling networks. We have recently developed new components and refined the existing ones to keep up with the computational demand of increasingly large datasets, including (i) an efficient integer linear programming, (ii) a probabilistic logic implementation for semi-quantitative datasets, (iii) the integration of a stochastic Boolean simulator, (iv) a tool to identify missing links, (v) systematic post-hoc analyses and (vi) an R-Shiny tool to run CellNOpt interactively. AVAILABILITY AND IMPLEMENTATION: R-package(s): https://github.com/saezlab/cellnopt. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: In silico drug target prediction provides valuable information for drug repurposing, understanding of side effects as well as expansion of the druggable genome. In particular, discovery of actionable drug targets is critical to developing targeted therapies for diseases. RESULTS: Here, we develop a robust method for drug target prediction by leveraging a class imbalance-tolerant machine learning framework with a novel training scheme. We incorporate novel features, including drug-gene phenotype similarity and gene expression profile similarity that capture information orthogonal to other features. We show that our classifier achieves robust performance and is able to predict gene targets for new drugs as well as drugs that potentially target unexplored genes. By providing newly predicted drug-target associations, we uncover novel opportunities of drug repurposing that may benefit cancer treatment through action on either known drug targets or currently undrugged genes. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SUMMARY: Visualizing a network provides a concise and practical understanding of the information it represents. Open-source web-based libraries help accelerate the creation of biologically based networks and their use. ccNetViz is an open-source, high speed and lightweight JavaScript library for visualization of large and complex networks. It implements customization and analytical features for easy network interpretation. These features include edge and node animations, which illustrate the flow of information through a network as well as node statistics. Properties can be defined a priori or dynamically imported from models and simulations. ccNetViz is thus a network visualization library particularly suited for systems biology. AVAILABILITY AND IMPLEMENTATION: The ccNetViz library, demos and documentation are freely available at http://helikarlab.github.io/ccNetViz/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The 16th Annual MidSouth Computational Biology and Bioinformatics Society (MCBIOS XVI) conference was held in Hilton Birmingham at University of Alabama Birmingham (UAB) conference center on March 28–30, 2019. The theme of the conference was Informatics for Precision Medicine. The co-chairs and conference hosts were Drs. Jake Y. Chen and Matthew Might from the University of Alabama Birmingham.

A case was recently brought to the journal’s attention regarding a reviewer who had requested a large number of citations to their own papers as part of their review. After investigation of their most recent reviews, we found that in every review this reviewer requested an average of 35 citations be added, ∼90% of which were to their own papers and the remainder to papers that both cited them extensively and mentioned them by name in the title. The reviewer’s phrasing strongly suggested that inclusion of these citations would influence their recommendation to the editor to accept or reject the paper. The reviewer was unable to provide a satisfactory justification for these requests and Bioinformatics has therefore banned them as a reviewer. Our investigation also suggests that the reviewer has behaved similarly in reviewing for other journals. This case has alerted us to how the peer-review system is vulnerable to unethical behavior, and prompted us to clarify the journal’s policy on when it is appropriate for reviewers to request citations to their own work, and to suggest how some of the current weak points in the peer-review system can be mitigated, so that this behavior can be detected more quickly and efficiently. Peer-reviewers are typically selected based on their expertise in the areas of research associated with newly submitted manuscripts. They are among the most likely to be familiar with prior publications pertinent to the submission, and reviewer feedback on the completeness and accuracy of a manuscript’s reference list is desirable and welcome. It is therefore not unusual that one or a few requested citations may be to the reviewer’s own research. However, reviewers should be aware that a rigorous scientific justification for the inclusion of a new citation must be provided. Since it is easy to provide a tenuous justification for inclusion, as this reviewer often did (e.g. that his papers also involved analysis of sequence data), it should instead be stated why the authors would be remiss or the paper weaker if the citation were not included. Likewise, editors and authors should be aware of the imbalance of power that exists in the review process, and should ensure that any citations added in response to a reviewer comment are relevant and important. Citations have been called the ‘currency’ of science, meaning that they could be considered a quantifiable and objective metric of the impact of a scientist’s research. Specific metrics, such as the H-index, are intended to reflect this and scientists therefore have an incentive to try to improve their H-index. Indeed, the reviewer we caught requesting extensive citation of their work has a webpage that includes prominent mentions of both their high H-index and past awards they received from Thomson Reuters for being a highly cited researcher. When a reviewer agrees to review a paper with the intention of inflating the number of citations to their work, this is a conflict of interest and an unethical manipulation of the peer-review system. One might ask how this reviewer got away with submitting multiple reviews containing coercive requests for citation before being banned. The shortest explanation is that excessive self-citation demands are generally not seen as an ethical problem until a pattern is established, and a decentralized peer-review system is not amenable to detecting patterns. Editors may overlook requests, authors seem reluctant to bring it up explicitly to the editor, reviewer comments are anonymous and scattered across different editors and different journals, and even editors that spot such patterns may not even be aware what options they have, and therefore take the least-energy option of no longer inviting that reviewer. Because accusing someone of unethical conduct is a serious matter, the editors and authors involved are hesitant to do so, particularly if all they have is one instance to base their actions upon. Combined, this creates a system whereby such behavior can persist for a very long time. Even in the rare event that such misbehavior is detected, there is no global solution. Following the guidelines of the Committee on Publication Ethics (COPE), the first step is to contact the reviewer for an explanation. If it is unsatisfactory, then bring the matter to the attention of their immediate supervisor. This second step is only effective in a well-organized academic system, which was not the case here. Despite the actions taken by Bioinformatics, it is likely that this reviewer will continue to review for other journals after this editorial is published. Readers have no doubt noticed that this reviewer is not mentioned by name. This is because we have concluded that we have an obligation to maintain peer-reviewer anonymity, and thus, we can only alert others to the general problem. We debated this last point extensively, and have raised the issue with the COPE as a case report to be discussed. Specific suggestions for reviewers: Motivation. Reviewers should properly motivate their requests for citations and specify how strongly they feel about the addition of references. For example, saying ‘the authors’ review of the field is incomplete, they should add the following references’ is vague, whereas ‘similar studies on the use of X for the purpose of Y were published prior to this one and are needed to alert readers to prior art’ is specific. The less specific a reviewer is regarding motivation, the less weight their request should be given. Moderation. Reviewers should refrain from requesting substantial numbers of references. What is ‘substantial’ will vary by the type of article, with review articles expected to be better in their coverage and short two-page application notes expected to include only the most relevant references. We propose a general rule of thumb to define ‘substantial’ as requesting addition of more than one reference per printed journal page of the paper. In the event the authors’ citations of pertinent prior research is highly incomplete, a reviewer should simply say so and then point them in the right direction with a few citations and let them do the rest. Communication. If a reviewer notices another reviewer has requested excessive or unmerited citations and this has not been commented on by the editor, they should feel free to share their observations and opinions directly with the editor. Reviewers should be cognizant they are also in a position to recognize patterns of abuse from their fellow reviewers. Specific suggestions for journals: Document patterns. Manuscript handling systems should include a checkbox for each reviewer that asks ‘did this reviewer request citation to their own research?’ Editors with a concern about reviewer’s citation requests could then see what percentage of reviews returned contained self-citation requests. Brief but clear guidelines. Instructions should be kept simple and clear. As a result of this case we have updated our reviewer guidelines to state that requests for citations should include ‘a brief, yet specific, rationale as to why their inclusion is merited. This rationale is particularly important if the reviewer requests citation of their own papers.’ Specific suggestions for authors: Voice concerns. Although adding multiple references in response to a reviewer request might seem like an ‘easy’ way to satisfy at least one of the reviewers, each unmerited citation clutters your paper and rewards unethical behavior. Don’t be hesitant to include in your response that you have considered the suggestion and feel they are not necessary. In the event the reviewer responds negatively, you should contact the editor for guidance. This is in accordance with the Ethical Guidelines for Reviewers published by the Committee on Publication Ethics (COPE Council. Ethical guidelines for peer reviewers. September 2017. www.publicationethics.org) Specific suggestions for editors handling papers: Vigilance. The ultimate responsibility in preventing this behavior lies with the editors handling the papers. Careful consideration of the referee reports to detect unethical behavior, including unjustified requests for citations, particularly citations of the reviewer’s own work, is important and all efforts should be made to prevent such requests being made to the authors. Importantly, when a reviewer requests substantial self-citation, this should be reported to the journal so they can investigate whether or not this is part of a pattern, as in the specific case discussed here. Bioinformatics acknowledges that their editorial controls have failed for some time in this particular case, and sincerely apologizes to our authors, referees and readers for not detecting this sooner. This phenomenon of reviewer-coerced citations is not new (Huggett, 2013; Ioannidis, 2015; Resnik et al., 2008; Thombs and Razykov, 2012; Thombs et al., 2015; Wilhite and Fong, 2012), but also not very well explored in terms of how extensive it may be or how it should be dealt with. We hope this editorial will prompt some discussion on the appropriate balance between the need for peer-reviewer anonymity and the need to alert others to potentially unethical behavior once a pattern is established, particularly when it is difficult to detect such patterns. It is possible that eliminating some of the anonymity, either by open peer-review or publishing anonymized peer reviews alongside accepted papers may disincentivize this behavior. Similarly, because highly centralized research resources, such as Publons and ORCID, have been developed, we hope that some ideas or discussion could take place regarding how these or similar centralized resources could be used, responsibly, to help document patterns of ethical concern that are otherwise difficult to detect. Conflict of Interest: none declared.

BACKGROUND: The number of publicly available metagenomic experiments in various environments has been rapidly growing, empowering the potential to identify similar shifts in species abundance between different experiments. This could be a potentially powerful way to interpret new experiments, by identifying common themes and causes behind changes in species abundance. RESULTS: We propose a novel framework for comparing microbial shifts between conditions. Using data from one of the largest human metagenome projects to date, the American Gut Project (AGP), we obtain differential abundance vectors for microbes using experimental condition information provided with the AGP metadata, such as patient age, dietary habits, or health status. We show it can be used to identify similar and opposing shifts in microbial species, and infer putative interactions between microbes. Our results show that groups of shifts with similar effects on microbiome can be identified and that similar dietary interventions display similar microbial abundance shifts. CONCLUSIONS: Without comparison to prior data, it is difficult for experimentalists to know if their observed changes in species abundance have been observed by others, both in their conditions and in others they would never consider comparable. Yet, this can be a very important contextual factor in interpreting the significance of a shift. We've proposed and tested an algorithmic solution to this problem, which also allows for comparing the metagenomic signature shifts between conditions in the existing body of data.

The XVth Annual MidSouth Computational Biology and Bioinformatics Society (MCBIOS XV) conference was held in Starkville, MS from March 29–31, 2018 at Mississippi State University (MSU) within the Mill Conference Center. MSU had previously hosted the conference (MCBIOS VI) in 2009. The theme of MCBIOS XV was “Genomics and Big Data”. The co-chairs and conference hosts were Drs. Bindu Nanduri, Andy Perkins, and Daniel G. Peterson from MSU. The program was co-chaired by Dr. Shraddha Thakkar from the National Center for Toxicological Research (NCTR) within the US Food and Drug Administration (FDA), Dr. Mary Yang, from University of Arkansas at Little Rock (UALR) and Dr. Prashanti Manda from University of North Carolina at Greensboro.

Motivation: Studies, mostly from the operations/management literature, have shown that the rate of human error increases with task complexity. What is not known is how many errors make it into the published literature, given that they must slip by peer-review. By identifying paired, dependent values within text for reported calculations of varying complexity, we can identify discrepancies, quantify error rates and identify mitigating factors. Results: We extracted statistical ratios from MEDLINE abstracts (hazard ratio, odds ratio, relative risk), their 95% CIs, and their P-values. We re-calculated the ratios and P-values using the reported CIs. For comparison, we also extracted percent-ratio pairs, one of the simplest calculation tasks. Over 486 000 published values were found and analyzed for discrepancies, allowing for rounding and significant figures. Per reported item, discrepancies were less frequent in percent-ratio calculations (2.7%) than in ratio-CI and P-value calculations (5.6-7.5%), and smaller discrepancies were more frequent than large ones. Systematic discrepancies (multiple incorrect calculations of the same type) were higher for more complex tasks (14.3%) than simple ones (6.7%). Discrepancy rates decreased with increasing journal impact factor (JIF) and increasing number of authors, but with diminishing returns and JIF accounting for most of the effect. Approximately 87% of the 81 937 extracted P-values were ≤ 0.05. Conclusion: Using a simple, yet accurate, approach to identifying paired values within text, we offer the first quantitative evaluation of published error frequencies within these types of calculations. Contact: [email protected] or [email protected]. Supplementary information: Supplementary data are available at Bioinformatics online.

BACKGROUND: NCBI's Gene Expression Omnibus (GEO) is a rich community resource containing millions of gene expression experiments from human, mouse, rat, and other model organisms. However, information about each experiment (metadata) is in the format of an open-ended, non-standardized textual description provided by the depositor. Thus, classification of experiments for meta-analysis by factors such as gender, age of the sample donor, and tissue of origin is not feasible without assigning labels to the experiments. Automated approaches are preferable for this, primarily because of the size and volume of the data to be processed, but also because it ensures standardization and consistency. While some of these labels can be extracted directly from the textual metadata, many of the data available do not contain explicit text informing the researcher about the age and gender of the subjects with the study. To bridge this gap, machine-learning methods can be trained to use the gene expression patterns associated with the text-derived labels to refine label-prediction confidence. RESULTS: Our analysis shows only 26% of metadata text contains information about gender and 21% about age. In order to ameliorate the lack of available labels for these data sets, we first extract labels from the textual metadata for each GEO RNA dataset and evaluate the performance against a gold standard of manually curated labels. We then use machine-learning methods to predict labels, based upon gene expression of the samples and compare this to the text-based method. CONCLUSION: Here we present an automated method to extract labels for age, gender, and tissue from textual metadata and GEO data using both a heuristic approach as well as machine learning. We show the two methods together improve accuracy of label assignment to GEO samples.

The XIVth Annual MidSouth Computational Biology and Bioinformatics Society was held in Little Rock, AR From March 23-25st 2017 and was co-hosted by University of Arkansas at Little Rock, University of Arkansas for Medical Sciences, Little Rock, AR and National Center for Toxicological Research, Jefferson, AR. The fourteenth annual conference entitled “ Make them Safer Make them Better: Bioinformatics and the Development of Therapeuticals ”. There were 220 conference registrants and 129 abstracts submitted, including 62 oral and 67 poster presentations.

MOTIVATION: The growing amount of regulatory data from the ENCODE, Roadmap Epigenomics and other consortia provides a wealth of opportunities to investigate the functional impact of single nucleotide polymorphisms (SNPs). Yet, given the large number of regulatory datasets, researchers are posed with a challenge of how to efficiently utilize them to interpret the functional impact of SNP sets. RESULTS: We developed the GenomeRunner web server to automate systematic statistical analysis of SNP sets within a regulatory context. Besides defining the functional impact of SNP sets, GenomeRunner implements novel regulatory similarity/differential analyses, and cell type-specific regulatory enrichment analysis. Validated against literature- and disease ontology-based approaches, analysis of 39 disease/trait-associated SNP sets demonstrated that the functional impact of SNP sets corresponds to known disease relationships. We identified a group of autoimmune diseases with SNPs distinctly enriched in the enhancers of T helper cell subpopulations, and demonstrated relevant cell type-specificity of the functional impact of other SNP sets. In summary, we show how systematic analysis of genomic data within a regulatory context can help interpreting the functional impact of SNP sets. AVAILABILITY AND IMPLEMENTATION: GenomeRunner web server is freely available at http://www.integrativegenomics.org/ CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: To analyze the relative proportion of bioinformatics papers and their non-bioinformatics counterparts in the top 20 most cited papers annually for the past two decades. RESULTS: When defining bioinformatics papers as encompassing both those that provide software for data analysis or methods underlying data analysis software, we find that over the past two decades, more than a third (34%) of the most cited papers in science were bioinformatics papers, which is approximately a 31-fold enrichment relative to the total number of bioinformatics papers published. More than half of the most cited papers during this span were bioinformatics papers. Yet, the average 5-year JIF of top 20 bioinformatics papers was 7.7, whereas the average JIF for top 20 non-bioinformatics papers was 25.8, significantly higher (P < 4.5 × 10(-29)). The 20-year trend in the average JIF between the two groups suggests the gap does not appear to be significantly narrowing. For a sampling of the journals producing top papers, bioinformatics journals tended to have higher Gini coefficients, suggesting that development of novel bioinformatics resources may be somewhat 'hit or miss'. That is, relative to other fields, bioinformatics produces some programs that are extremely widely adopted and cited, yet there are fewer of intermediate success. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The MidSouth Computational Biology and Bioinformatics Society (MCBIOS) held its thirteenth annual conference themed “Precision Medicine and Data Sciences” at the University of Memphis, FedEx Institute of Technology in Memphis, Tennessee on March 3–5, 2016. There were 156 conference registrants and 117 abstracts submitted, including 63 oral and 54 poster presentations.

BACKGROUND: Adapter trimming and removal of duplicate reads are common practices in next-generation sequencing pipelines. Sequencing reads ambiguously mapped to repetitive and low complexity regions can also be problematic for accurate assessment of the biological signal, yet their impact on sequencing data has not received much attention. We investigate how trimming the adapters, removing duplicates, and filtering out reads overlapping low complexity regions influence the significance of biological signal in RNA- and ChIP-seq experiments. METHODS: We assessed the effect of data processing steps on the alignment statistics and the functional enrichment analysis results of RNA- and ChIP-seq data. We compared differentially processed RNA-seq data with matching microarray data on the same patient samples to determine whether changes in pre-processing improved correlation between the two. We have developed a simple tool to remove low complexity regions, RepeatSoaker, available at https://github.com/mdozmorov/RepeatSoaker, and tested its effect on the alignment statistics and the results of the enrichment analyses. RESULTS: Both adapter trimming and duplicate removal moderately improved the strength of biological signals in RNA-seq and ChIP-seq data. Aggressive filtering of reads overlapping with low complexity regions, as defined by RepeatMasker, further improved the strength of biological signals, and the correlation between RNA-seq and microarray gene expression data. CONCLUSIONS: Adapter trimming and duplicates removal, coupled with filtering out reads overlapping low complexity regions, is shown to increase the quality and reliability of detecting biological signals in RNA-seq and ChIP-seq data.

BACKGROUND: As the amount of scientific data grows, peer-reviewed Scientific Data Analysis Resources (SDARs) such as published software programs, databases and web servers have had a strong impact on the productivity of scientific research. SDARs are typically linked to using an Internet URL, which have been shown to decay in a time-dependent fashion. What is less clear is whether or not SDAR-producing group size or prior experience in SDAR production correlates with SDAR persistence or whether certain institutions or regions account for a disproportionate number of peer-reviewed resources. METHODS: We first quantified the current availability of over 26,000 unique URLs published in MEDLINE abstracts/titles over the past 20 years, then extracted authorship, institutional and ZIP code data. We estimated which URLs were SDARs by using keyword proximity analysis. RESULTS: We identified 23,820 non-archival URLs produced between 1996 and 2013, out of which 11,977 were classified as SDARs. Production of SDARs as measured with the Gini coefficient is more widely distributed among institutions (.62) and ZIP codes (.65) than scientific research in general, which tends to be disproportionately clustered within elite institutions (.91) and ZIPs (.96). An estimated one percent of institutions produced 68% of published research whereas the top 1% only accounted for 16% of SDARs. Some labs produced many SDARs (maximum detected = 64), but 74% of SDAR-producing authors have only published one SDAR. Interestingly, decayed SDARs have significantly fewer average authors (4.33 +/- 3.06), than available SDARs (4.88 +/- 3.59) (p < 8.32 × 10-4). Approximately 3.4% of URLs, as published, contain errors in their entry/format, including DOIs and links to clinical trials registry numbers. CONCLUSION: SDAR production is less dependent upon institutional location and resources, and SDAR online persistence does not seem to be a function of infrastructure or expertise. Yet, SDAR team size correlates positively with SDAR accessibility, suggesting a possible sociological factor involved. While a detectable URL entry error rate of 3.4% is relatively low, it raises the question of whether or not this is a general error rate that extends to additional published entities.

The MidSouth Computational Biology and Bioinformatics Society (MCBIOS 2014) held its eleventh annual conference at the Wes Watkins Center at Oklahoma State University, Stillwater on March 7-8, 2014. The theme was " From Genome to Phenome: Connecting the Dots ". Conference Chair this year was Rakesh Kaundal, who is also one of the MCBIOS board members, and conference committee members were Ulrich K. Melcher and Doris Kupfer. The current president is Andy Perkins and Cesar Compadre was elected as President-Elect for 2015-16. There were 154 registrants and a total of 125 abstracts submitted (50 oral and 75 poster presentations).

Even though only 1.5% of the human genome is translated into proteins, recent reports indicate that most of it is transcribed into non-coding RNAs (ncRNAs), which are becoming the subject of increased scientific interest. We hypothesized that examining how different classes of ncRNAs co-localized with annotated epigenomic elements could help understand the functions, regulatory mechanisms, and relationships among ncRNA families. We examined 15 different ncRNA classes for statistically significant genomic co-localizations with cell type-specific chromatin segmentation states, transcription factor binding sites (TFBSs), and histone modification marks using GenomeRunner ( http://www.genomerunner.org ). P-values were obtained using a Chi-square test and corrected for multiple testing using the Benjamini-Hochberg procedure. We clustered and visualized the ncRNA classes by the strength of their statistical enrichments and depletions. We found piwi-interacting RNAs (piRNAs) to be depleted in regions containing activating histone modification marks, such as H3K4 mono-, di- and trimethylation, H3K27 acetylation, as well as certain TFBSs. piRNAs were further depleted in active promoters, weak transcription, and transcription elongation regions, and enriched in repressed and heterochromatic regions. Conversely, transfer RNAs (tRNAs) were depleted in heterochromatin regions and strongly enriched in regions containing activating H3K4 di- and trimethylation marks, H2az histone variant, and a variety of TFBSs. Interestingly, regions containing CTCF insulator protein binding sites were associated with tRNAs. tRNAs were also enriched in the active, weak and poised promoters and, surprisingly, in regions with repetitive/copy number variations. Searching for statistically significant associations between ncRNA classes and epigenomic elements permits detection of potential functional and/or regulatory relationships among ncRNA classes, and suggests cell type-specific biological roles of ncRNAs.

BACKGROUND: Experimentally validated co-expression correlations between miRNAs and genes are a valuable resource to corroborate observations about miRNA/mRNA changes after experimental perturbations, as well as compare miRNA target predictions with empirical observations. For example, when a given miRNA is transcribed, true targets of that miRNA should tend to have lower expression levels relative to when the miRNA is not expressed. METHODS: We processed publicly available human RNA-seq experiments obtained from NCBI's Sequence Read Archive (SRA) to identify miRNA-mRNA co-expression trends and summarized them in terms of their Pearson's Correlation Coefficient (PCC) and significance. RESULTS: We found that sequence-derived parameters from TargetScan and miRanda were predictive of co-expression, and that TargetScan- and miRanda-derived gene-miRNA pairs tend to have anti-correlated expression patterns in RNA-seq data compared to controls. We provide this data for download and as a web application available at http://wrenlab.org/mirCoX/. CONCLUSION: This database of empirically established miRNA-mRNA transcriptional correlations will help to corroborate experimental observations and could be used to help refine and validate miRNA target predictions.

The tenth annual conference of the MidSouth Computational Biology and Bioinformatics Society (MCBIOS 2013), "The 10th Anniversary in a Decade of Change: Discovery in a Sea of Data", took place at the Stoney Creek Inn & Conference Center in Columbia, Missouri on April 5-6, 2013. This year's Conference Chairs were Gordon Springer and Chi-Ren Shyu from the University of Missouri and Edward Perkins from the US Army Corps of Engineers Engineering Research and Development Center, who is also the current MCBIOS President (2012-3). There were 151 registrants and a total of 111 abstracts (51 oral presentations and 60 poster session abstracts).

MOTIVATION: One of the challenges in interpreting high-throughput genomic studies such as a genome-wide associations, microarray or ChIP-seq is their open-ended nature-once a set of experimentally identified regions is identified as statistically significant, at least two questions arise: (i) besides P-value, do any of these significant regions stand out in terms of biological implications? (ii) Does the set of significant regions, as a whole, have anything in common genome wide? These issues are difficult to address because of the growing number of annotated genomic features (e.g. single nucleotide polymorphisms, transcription factor binding sites, methylation peaks, etc.), and it is difficult to know a priori which features would be most fruitful to analyze. Our goal is to provide partial automation of this process to begin examining associations between experimental features and annotated genomic regions in a hypothesis-free, data-driven manner. RESULTS: We created GenomeRunner-a tool for automating annotation and enrichment of genomic features of interest (FOI) with annotated genomic features (GFs), in different organisms. Besides simple association of FOIs with known GFs GenomeRunner tests whether the enriched FOIs, as a group, are statistically associated with a large and growing set of genomic features. AVAILABILITY: GenomeRunner setup files and source code are freely available at http://sourceforge.net/projects/genomerunner. CONTACT: [email protected]; [email protected]; [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The last two decades have witnessed great advances in biological tissue labeling and automated microscopic imaging that, in turn, have revolutionized how biologists visualize molecular, sub-cellular, cellular, and super-cellular structures and study their respective functions. Tremendous volumes of multi-dimensional bioimaging data are now being generated in almost every branch of biology. How to interpret such image datasets in a quantitative, objective, automatic and efficient way has become a major challenge in current computational biology. Bioimage informatics methods have begun to turn image data into useful biological knowledge (Peng, 2008; Swedlow, et al., 2009; Shamir, et al., 2010; Danuser, 2011). The essential methods of bioimage informatics involve large-scale bioimage generation, visualization, analysis and management. Bioimage informatics also encompasses both hypothesis- and data-driven exploratory approaches, with an emphasis on how to generate biological knowledge and/or gain new insights that would otherwise be hard to achieve. Early work in bioimage informatics began in the late 1990s. Increasingly, computer vision, image analysis, data mining, machine learning and pattern recognition methods have been applied to microscopic images to extract biological information and to generate ontology databases. The growing amount of bioimage data are quickly imposing additional demands on how to store, manage and retrieve such image datasets as well as the associated secondary meta-data. Data analysis, fusion and reconstruction techniques have also been developed to facilitate better image acquisition and formation. Joint analysis of image data in combination with other biological datasets, such as genomes and gene expression profiles, is also becoming more and more commonplace. To meet the need of this growing field, the first international workshop on Bioimage Informatics was organized at Stanford University in 2005. It grew to be an annual event in this field. Other meetings on similar topics and related applications have also emerged since then. In 2010, the annual conference on Intelligent Systems for Molecular Biology (ISMB) established a paper-submission track on bioimaging data analysis and visualization. While there is a noticeable need to publish high quality papers on bioimage informatics, so far no high-impact journal explicitly accepts this category of papers. We believe it is an appropriate time to create this new category in Bioinformatics. As of February 2012, Bioinformatics now includes a new paper submission category in the scope described by the journal at its website as follows: ‘Informatics methods for the acquisition, analysis, mining and visualization of images produced by modern microscopy, with an emphasis on the application of novel computing techniques to solve challenging and significant biological and medical problems at the molecular, sub-cellular, cellular, and super-cellular (organ, organism, and population) levels. This category also encourages large-scale image informatics methods/applications/software, various enabling techniques (e.g. cyber-infrastructures, quantitative validation experiments, pattern recognition, etc.) for such large-scale studies, and joint analysis of multiple heterogeneous datasets that include images as a component. Bioimage related ontology and databases studies, image-oriented large-scale machine learning, data mining, and other analytics techniques are also encouraged. We will not consider image analysis and pattern recognition methods that are solely based on tuning parameters or swapping computational sub-steps, without an in-depth description or demonstration of why such changes are significantly superior for one or more biological problems.’ We would like to thank a number of colleagues and practitioners who have contributed to the creation of this new category. Especially, we thank Eugene Myers, B.S.Manjunath, Badri Roysam, Manfred Auer, Michael Hawrylycz, Jean-Christophe Olivo-Marin, Anne Carpenter and Vebjorn Ljosa in helping to define this new category of paper submissions. We hope the journal Bioinformatics becomes a valuable venue for bioimage informatics researchers to publish their most important work. Contact: gro.imhh.ailenaj@hgnep

The ninth annual conference of the MidSouth Computational Biology and Bioinformatics Society (MCBIOS 2012), "Making Sense of the Omics Data Deluge", took place in Oxford, Mississippi February 17-8 2012. This year's Conference Chairs were Dr. Dawn Wilkins, of the University of Mississippi and Dr. Doris Kupfer, also the current MCBIOS President (2011-2), from the Federal Aviation Administration. There were 170 registrants and a total of 106 abstracts (34 oral presentations and 72 poster session abstracts).

Abstract Contact: [email protected]

BACKGROUND: Global meta-analysis (GMA) of microarray data to identify genes with highly similar co-expression profiles is emerging as an accurate method to predict gene function and phenotype, even in the absence of published data on the gene(s) being analyzed. With a third of human genes still uncharacterized, this approach is a promising way to direct experiments and rapidly understand the biological roles of genes. To predict function for genes of interest, GMA relies on a guilt-by-association approach to identify sets of genes with known functions that are consistently co-expressed with it across different experimental conditions, suggesting coordinated regulation for a specific biological purpose. Our goal here is to define how sample, dataset size and ranking parameters affect prediction performance. RESULTS: 13,000 human 1-color microarrays were downloaded from GEO for GMA analysis. Prediction performance was benchmarked by calculating the distance within the Gene Ontology (GO) tree between predicted function and annotated function for sets of 100 randomly selected genes. We find the number of new predicted functions rises as more datasets are added, but begins to saturate at a sample size of approximately 2,000 experiments. For the gene set used to predict function, we find precision to be higher with smaller set sizes, yet with correspondingly poor recall and, as set size is increased, recall and F-measure also tend to increase but at the cost of precision. CONCLUSIONS: Of the 20,813 genes expressed in 50 or more experiments, at least one predicted GO category was found for 72.5% of them. Of the 5,720 genes without GO annotation, 4,189 had at least one predicted ontology using top 40 co-expressed genes for prediction analysis. For the remaining 1,531 genes without GO predictions or annotations, ~17% (257 genes) had sufficient co-expression data yet no statistically significantly overrepresented ontologies, suggesting their regulation may be more complex.

BACKGROUND: Microarray experiments are becoming increasingly common in biomedical research, as is their deposition in publicly accessible repositories, such as Gene Expression Omnibus (GEO). As such, there has been a surge in interest to use this microarray data for meta-analytic approaches, whether to increase sample size for a more powerful analysis of a specific disease (e.g. lung cancer) or to re-examine experiments for reasons different than those examined in the initial, publishing study that generated them. For the average biomedical researcher, there are a number of practical barriers to conducting such meta-analyses such as manually aggregating, filtering and formatting the data. Methods to automatically process large repositories of microarray data into a standardized, directly comparable format will enable easier and more reliable access to microarray data to conduct meta-analyses. METHODS: We present a straightforward, simple but robust against potential outliers method for automatic quality control and pre-processing of tens of thousands of single-channel microarray data files. GEO GDS files are quality checked by comparing parametric distributions and quantile normalized to enable direct comparison of expression level for subsequent meta-analyses. RESULTS: 13,000 human 1-color experiments were processed to create a single gene expression matrix that subsets can be extracted from to conduct meta-analyses. Interestingly, we found that when conducting a global meta-analysis of gene-gene co-expression patterns across all 13,000 experiments to predict gene function, normalization had minimal improvement over using the raw data. CONCLUSIONS: Normalization of microarray data appears to be of minimal importance on analyses based on co-expression patterns when the sample size is on the order of thousands microarray datasets. Smaller subsets, however, are more prone to aberrations and artefacts, and effective means of automating normalization procedures not only empowers meta-analytic approaches, but aids in reproducibility by providing a standard way of approaching the problem.Data availability: matrix containing normalized expression of 20,813 genes across 13,000 experiments is available for download at . Source code for GDS files pre-processing is available from the authors upon request.

The Eighth Annual Conference of the MidSouth Computational Biology and Bioinformatics Society (MCBIOS’2011) was held in College Station, Texas on April 1-2, 2011. The Conference General Chair was Ulisses Braga-Neto, the MCBIOS President for the 2010-2011 term, from Texas A&M University. There were nearly 200 registrants and 140 abstracts were submitted, divided into 48 oral presentation abstracts and 92 poster session abstracts.

MOTIVATION: Document similarity metrics such as PubMed's 'Find related articles' feature, which have been primarily used to identify studies with similar topics, can now also be used to detect duplicated or potentially plagiarized papers within literature reference databases. However, the CPU-intensive nature of document comparison has limited MEDLINE text similarity studies to the comparison of abstracts, which constitute only a small fraction of a publication's total text. Extending searches to include text archived by online search engines would drastically increase comparison ability. For large-scale studies, submitting short phrases encased in direct quotes to search engines for exact matches would be optimal for both individual queries and programmatic interfaces. We have derived a method of analyzing statistically improbable phrases (SIPs) for assistance in identifying duplicate content. RESULTS: When applied to MEDLINE citations, this method substantially improves upon previous algorithms in the detection of duplication citations, yielding a precision and recall of 78.9% (versus 50.3% for eTBLAST) and 99.6% (versus 99.8% for eTBLAST), respectively. AVAILABILITY: Similar citations identified by this work are freely accessible in the Déjà vu database, under the SIP discovery method category at http://dejavu.vbi.vt.edu/dejavu/.

The seventh annual Midsouth Computational Biology and Bioinformatics (MCBIOS) conference took place February 19 and 20, 2010 at Arkansas State University in Jonesboro, AR, presided over by Daniel Berleant, this year’s President of MCBIOS. Keynote speakers were Elaine Ostrander of NIH, renowned for her work on dog genomics, Clayton Naeve, CIO of St. Jude Children's Research Hospital, and Robert Cottingham, Leader of the Computational Biology and Bioinformatics Group, Oak Ridge National Laboratory. The record-breaking attendance exceeded 200 participants which necessitated parallel talk sessions for the first time. Student oral presentation award winners were Heidi Pagan (first place) of Mississippi State University (MSU), Juliet Tang (second place) of MSU, and Aleksandra Markovets (third) of Mississippi Valley State University. In addition, a record number of posters were also presented. Due to the number of posters, student poster awards were given in two categories, Biological Focus and Computational Focus. Winners were N. Platt and V. Chaitankar (1st place), G. Cooper and L. Pillai (2nd), and M. Ammari and C. Gearheart (3rd). Reflecting the integration of these foci in the field, winners in one category frequently scored high in the other category as well. MCBIOS is also pleased to have this year acquired legal status as a non-profit organization.

MOTIVATION: Approximately 9334 (37%) of human genes have no publications documenting their function and, for those that are published, the number of publications per gene is highly skewed. Furthermore, for reasons not clear, the entry of new gene names into the literature has slowed in recent years. If we are to better understand human/mammalian biology and complete the catalog of human gene function, it is important to finish predicting putative functions for these genes based upon existing experimental evidence. RESULTS: A global meta-analysis (GMA) of all publicly available GEO two-channel human microarray datasets (3551 experiments total) was conducted to identify genes with recurrent, reproducible patterns of co-regulation across different conditions. Patterns of co-expression were divided into parallel (i.e. genes are up and down-regulated together) and anti-parallel. Several ranking methods to predict a gene's function based on its top 20 co-expressed gene pairs were compared. In the best method, 34% of predicted Gene Ontology (GO) categories matched exactly with the known GO categories for approximately 5000 genes analyzed versus only 3% for random gene sets. Only 2.4% of co-expressed gene pairs were found as co-occurring gene pairs in MEDLINE. CONCLUSIONS: Via a GO enrichment analysis, genes co-expressed in parallel with the query gene were frequently associated with the same GO categories, whereas anti-parallel genes were not. Combining parallel and anti-parallel genes for analysis resulted in fewer significant GO categories, suggesting they are best analyzed separately. Expression databases contain much unexpected genetic knowledge that has not yet been reported in the literature. A total of 1642 Human genes with unknown function were differentially expressed in at least 30 experiments. AVAILABILITY: Data matrix available upon request.

MCBIOS 2009 was held February 20–21, 2009 at the Hunter Henry Center in Starkville, Mississippi on the Mississippi State University campus. The conference was hosted by the four research universities in Mississippi comprising the Mississippi Computational Biology Consortium – Jackson State University (JSU), Mississippi State University (MSU), the University of Mississippi (UM), and the University of Southern Mississippi (USM). Dr. Dawn Wilkins of UM and Dr. Susan Bridges of MSU served as conference co-chairs. Keynote speakers were Dr. Laura Elnitski, Head of the Genomic Functional Analysis Section at the National Human Genome Research Institute (NHGRI), Howard Cash, President and CEO of Gene Codes Corporation, and Dr. Cathy Wu, Director of the Protein Information Resource (PIR).

MOTIVATION: Duplicate publication impacts the quality of the scientific corpus, has been difficult to detect, and studies this far have been limited in scope and size. Using text similarity searches, we were able to identify signatures of duplicate citations among a body of abstracts. RESULTS: A sample of 62,213 Medline citations was examined and a database of manually verified duplicate citations was created to study author publication behavior. We found that 0.04% of the citations with no shared authors were highly similar and are thus potential cases of plagiarism. 1.35% with shared authors were sufficiently similar to be considered a duplicate. Extrapolating, this would correspond to 3500 and 117,500 duplicate citations in total, respectively. AVAILABILITY: eTBLAST, an automated citation matching tool, and Déjà vu, the duplicate citation database, are freely available at http://invention.swmed.edu/ and http://spore.swmed.edu/dejavu

MOTIVATION: Internet-based electronic resources, as given by Uniform Resource Locators (URLs), are being increasingly used in scientific publications but are also becoming inaccessible in a time-dependant manner, a phenomenon documented across disciplines. Initial reports brought attention to the problem, spawning methods of effectively preserving URL content while some journals adopted policies regarding URL publication and begun storing supplementary information on journal websites. Thus, a reexamination of URL growth and decay in the literature is merited to see if the problem has grown or been mitigated by any of these changes. RESULTS: After the 2003 study, three follow-up studies were conducted in 2004, 2005 and 2007. Unfortunately, no significant change was found in the rate of URL decay among any of the studies. However, only 5% of URLs cited more than twice have decayed versus 20% of URLs cited once or twice. The most common types of lost content were computer programs (43%), followed by scholarly content (38%) and databases (19%). Compared to URLs still available, no lost content type was significantly over- or underrepresented. Searching for 30 of these websites using Google, 11 (37%) were found relocated to different URLs. CONCLUSIONS: URL decay continues unabated, but URLs published by organizations tend to be more stable. Repeated citation of URLs suggests calculation of an electronic impact factor (eIF) would be an objective, quantitative way to measure the impact of Internet-based resources on scientific research.

UNLABELLED: As biomedical data accumulates, the need to store, share and organize it grows. Consequently, the number of Internet-accessible databases has been rapidly growing on an annual basis. Bioinformatics regularly publishes descriptions of biomedically relevant databases, Nucleic Acids Research has published an annual database issue since 1996 and now a new open-access journal, DATABASE: The Journal of Biological DATABASEs and Curation, will soon be launched by Oxford University Press in 2009 (http://www.oxfordjournals.org/our_journals/databa/). Since databases can be made publicly available on the Internet without publication, it is worth considering what factors prioritize publication of database descriptions in a peer-reviewed journal. In general, publication of a database description in a journal advertises it as a valuable resource for scientific research. Implicitly, it is assumed that this resource is publicly available (most likely for free) and will be maintained. However, therein lies the problem: DATABASE papers are simply not of the same nature as regular research articles. Over time, some databases simply become inaccessible, some are created but not maintained or updated, and some databases are never used (Galperin, 2006). Thus, for database creators, reviewers and journal editors, there are several additional considerations to judge, prior to publication, how potentially valuable these new databases may be.

BACKGROUND: Relationships between entities such as genes, chemicals, metabolites, phenotypes and diseases in MEDLINE are often directional. That is, one may affect the other in a positive or negative manner. Detection of causality and direction is key in piecing pathways together and in examining possible implications of experimental results. Because of the size and growth of biomedical literature, it is increasingly important to be able to automate this process as much as possible. RESULTS: Here we present a method of relation extraction using dependency graph parsing with SVM classification. We tested the SVM classifier first on gold standard corpora from GENIA and find it achieved 82% precision and 94.8% recall (F-measure of 87.9) on these standardized test sets. We then applied the entire system to all available MEDLINE abstracts for two target interactions with known effects. We find that while some directional relations are extracted with low ambiguity, others are apparently contradictory, at least when considered in an isolated context. When examined, it is apparent some are dependent upon the surrounding context (e.g. whether the relationship referred to short-term or long-term effects, or whether the focus was extracellular versus intracellular). CONCLUSION: Thesaurus-based directional relation extraction can be done with reasonable accuracy, but is prone to false-positives on larger corpora due to noun modifiers. Furthermore, methods of resolving or disambiguating relationship context and contingencies are important for large-scale corpora.

MCBIOS 2008 was held February 23–24, 2008 in Oklahoma City, Oklahoma at the Cox Convention Center in Bricktown. It was the best attended in the series of MCBIOS conferences (140 registrants) with the most participation (68 posters submitted). Informative and engaging keynote talks were delivered by Dr. Bruce Roe and Dr. Edward Dougherty. The full agenda is online at http://www.okbios.org .

The cross-disciplinary nature of bioinformatics entails co-evolution with other biomedical disciplines, whereby some bioinformatics applications become popular in certain disciplines and, in turn, these disciplines influence the focus of future bioinformatics development efforts. We observe here that the growth of computational approaches within various biomedical disciplines is not merely a reflection of a general extended usage of computers and the Internet, but due to the production of useful bioinformatics databases and methods for the rest of the biomedical scientific community. We have used the abstracts stored both in the MEDLINE database of biomedical literature and in NIH-funded project grants, to quantify two effects. First, we examine the biomedical literature as a whole and find that the use of computational methods has become increasingly prevalent across biomedical disciplines over the past three decades, while use of databases and the Internet have been rapidly increasing over the past decade. Second, we study the recent trends in the use of bioinformatics topics. We observe that molecular sequence databases are a widely adopted contribution in biomedicine from the field of bioinformatics, and that microarray analysis is one of the major new topics engaged by the bioinformatics community. Via this analysis, we were able to identify areas of rapid growth in the use of informatics to aid in curriculum planning, development of computational infrastructure and strategies for workforce education and funding.

MOTIVATION: Microarrays rapidly generate large quantities of gene expression information, but interpreting such data within a biological context is still relatively complex and laborious. New methods that can identify functionally related genes via shared literature concepts will be useful in addressing these needs. RESULTS: We have developed a novel method that uses implicit literature relationships (concepts related via shared, intermediate concepts) to cluster related genes. Genes are evaluated for implicit connections within a network of biomedical objects (other genes, ontological concepts and diseases) that are connected via their co-occurrences in Medline titles and/or abstracts. On the basis of these implicit relationships, individual gene pairs are scored using a probability-based algorithm. Scores are generated for all pairwise combinations of genes, which are then clustered based on the scores. We applied this method to a test set composed of nine functional groups with known relationships. The method scored highly for all nine groups and significantly better than a benchmark co-occurrence-based method for six groups. We then applied this method to gene sets specific to two previously defined breast tumor subtypes. Analysis of the results recapitulated known biological relationships and identified novel pathway relationships unique to each tumor subtype. We demonstrate that this method provides a valuable new means of identifying and visualizing significantly related genes within gene lists via their implicit relationships in the literature.

The Fourth Annual MidSouth Computational Biology and Bioinformatics Society (MCBIOS-IV) conference was held in New Orleans, Louisiana on February 1 st –3 rd , 2007. The conference venue included two locations: the Lindy Boggs Conference facility at the University of New Orleans (UNO); and Dinner/Speaker venues in the French Quarter – Broussard's of New Orleans and The House of Blues. The conference featured three days of technical presentations, with the third day partly devoted to a satellite Conference on Nanopore Cheminformatics with an on-site demo presentation of a nanopore detector (from the Research Institute for Children and UNO Nanopore Cheminformatics/Biophysics Labs). The theme of this year's conference was "Computational Frontiers in Biomedicine".

Are software patents bad for developers and researchers? The debate has been raging on this topic as more and more software patents are issued and intellectual property staked out, but the answer ultimately is not as simple as a yes or no. Patents arose as a legal mechanism designed to protect ‘intellectual capital’; that is, to protect the ideas of inventors from being used by others without permission and/or compensation. It was recognized early on that if one wished to provide incentive to innovators, that not all inventors had the means to implement, develop or even produce their invention at the time it was conceived. Without some kind of protection, inventors were reluctant to aggressively pursue their ideas, much less share them with others that might prefer stealing the idea to sharing the glory. And, as a general rule, people often scoff at new ideas until someone proves they are useful, which usually requires someone to invest significant resources such as time, money and effort. In theory, patents provide a valuable protection system for inventors pursuing development of their invention as well as a reward system for commercially valuable ideas. Modern-day patents, however, are not always of such a noble nature and software patents, a relatively new concept, are becoming more of an issue in bioinformatics (Lesavich, 2000). These patents may eventually wind up as landmines in the path of progress, as there is now clearly a divergence between theory and reality. The term ‘software patents’ is a little bit of a misnomer since in general, it is not the software that is patented, but the underlying method or algorithmic means of accomplishing an end result. It would not make much sense, for example, to patent your Java code only to have someone circumvent it by writing a slightly different program to do the same thing in C. Software patents are also known as ‘computer implemented inventions’. And there are two major areas whereby software patents pose a potential problem to bioinformatics research: first, part of the problem lies with the criteria by which patents are granted. To obtain a patent for a claimed invention, it is supposed to be new, useful and non-obvious. If just one of these criteria is missing, then—in theory—the patent application should be rejected. Software differs from most other inventions in that end-users never see what was done algorithmically to achieve the end program—programmers are constantly reinventing the wheel and rarely documenting what they did in a publicly searchable format that could be used to establish prior art. And unfortunately each of the criteria for patenting can be subjective, which is reflected in the patent database. For example, it has been pointed out (see Author Webpage) that at least five patents have been granted for the method of using a laser pointer to amuse a pet (Amiss and Abbott, 1993; Lorenz, 2001; Nathanson and Kersaint, 2001; Allen, 2002; Chelen, 2002). Four of these patents were granted despite the existence of previous patents of the same method and that of a publication predating even the earliest of these patents by over a decade describing a virtually identical method (Stidger and Stidger, 1982). Yet, until the patents are officially overturned, each of these inventors can legally sue infringers, including each other. Lawsuits, of course, cost time and money. While in some cases such as this one, they might be settled relatively quickly, in other cases they may drag on for a long time … which leads us to the second major problem with software patents. The second problem with software patents, and perhaps the patent system in general, is the issuance of broad and/or vague patents. Of course, the broader the claims, the more likely they are to either be considered obvious or to impinge upon prior art and be rejected at the application stage. Yet, ironically, despite the higher probability of being overturned, vague patents can be even more valuable than specific ones if they are ultimately issued, as they permit the inventor to subsequently lay claim to a broad swath of intellectual territory, including specific ideas they would never have conceived of independently. These patents can become serious stumbling blocks to progress, and the implications of this range from the mundane to the severe. Few people would care if a clever graphical interface never made it to market because of litigation, but what about an algorithm that helps to better understand, diagnose or treat a disease? Fortunately, at least, congress is currently considering a patent reform package that includes a provision to eliminate the injunction associated with patent infringement if the patent holders are not themselves making the product. These two problems, combined with an explosive growth in the number of software patents, have led to a third problem—defensive patenting. For most companies that actually produce a software product, the motivation is not so much to sue other people for infringing their patent, but to avoid being sued. Even if they can clearly show that they invented something prior to another patent claiming the same invention, litigation can still be costly. The original intent behind their patenting, however, may change—these patents can equally be used offensively. It has been convincingly argued that the biggest beneficiaries of software patents are lawyers. The tie-in to bioinformatics research should be apparent for software developers: How many patent ‘landmines’ might you be in danger of stepping on when you write your next program? And let us face it, if someone can patent a method for swinging on a swing or the use of a wheel for transportation (Olson, 2000; Hecht, 2002), it may just be a matter of time before someone tries to patent the idea of a For…Next loop (e.g. ‘We describe a method for the iterative algorithmic processing of command statements, the number of iterations being fully customizable and controllable by a user…’). Let us just hope the patent office is vigilant if they do. Yet, at the same time, it should be noted that there have been no high-profile cases of bioinformatics research being hindered by patent issues so far. In an odd way, the same shady motivations of software patenting and litigation that put companies at risk protect the academic enterprise. There is normally very little money in suing researchers, after all. So, although we may technically already be in violation of hundreds of patents from the software we have written and published, I am willing to bet most of us will never know about it unless we become filthy rich as a direct consequence. However, the potential threat software patents could pose to bioinformatics research should not be ignored, especially when it comes to the growing array of web services and downloadable analysis programs developed by researchers, which could cut into the sales of a company trying to provide a similar service. With all these potential problems, why not just end the practice of issuing software patents? Salzberg and Quackenbush argue that we should (Salzberg and Quackenbush, 2006). I would argue that software patents are not inherently evil, and that the potential for profit provides more motivation to pursue the realization of a powerful idea than academic considerations alone. Thus, software patents should not be abolished, but the system reformed. Consider an example: In 1998, Sergey Brin and Larry Page published a paper on a search engine they were working on called Google (Brin and Page, 1998). They filed a patent the same year (Page, 1998), which was granted in 2001. At the time, the reigning search engine giants like Yahoo, Altavista and Webcrawler were already well established, flush with resources—at least compared to an academic start-up that began in a garage. Google is today the world's most popular search engine, earning that distinction by giving people a superior product. But what would have happened to Google if their idea could have been stolen without legal consequence? First of all, I doubt that investors would have risked their money on an idea that could not be protected but, assuming they did, then I suspect the reigning search engines would have ignored Google when it began. Why change their apparently successful approaches, after all? But around 2000 or so, after a lot of resources by the Google team had already been committed and their success was growing rapidly, I think the same search engines would have begun re-evaluating the situation and considered adopting Google's methods in face of this growing threat. They could have used their existing resources to re-implement Brin and Page's algorithm (thereby avoiding copyright issues) and quash this newcomer before it could grow too large and powerful. And we, the consumers, would likely be worse off, since the value Google offers as a company encompasses many other services besides their search engine (e.g. gmail, free APIs). Thus, if software patents can be used to protect inventors and investors actively pursuing development of a clever idea, then they can be a good thing. So the question remains of how to keep the potential for good and lessen the potential for evil. First, the criteria for awarding patents should include the intent to pursue development of an invention. Some might argue that it is ok to conceive of an idea, patent it and wait to sue for infringement anyone who actually develops a working system that was only an abstract concept in their patent application. Euphemistically, this would be termed a ‘business strategy’ or ‘strategic intellectual property placement’. I would argue that, while legal, this is completely contrary to the original intents behind patent law. This ‘patent trolling’ is a travesty that has the potential to completely hinder progress for the 20 year lifetime of the patent. And, until recently, another abuse of the (US) patent system enabled progress to be stymied for even longer—a tactic known as ‘submarine patents’ (see Author Webpage) permitted patent filers to continually revise their application and protect their IP while postponing the actual start date. This was recently abolished by changing patent life from 17 years from issuance to 20 years from initial filing. But this is still a long time to ‘sit’ on an idea. Patent life should be shortened to 5 year terms, with 4 consecutive renewals possible. Each renewal would be contingent upon showing tangible or continued progress in pursuing the idea. People who are actually developing can easily show it, but companies that want to sit on a large patent portfolio will get overwhelmed quickly if they have to constantly pretend they are working. Their former ‘assets’ will become liabilities. Currently, there are patent maintenance fees that must be paid at 3.5 years ($450), 7.5 years ($1150) and 11.5 years ($1900), and this helps slightly. But given that patent-holders pay between $3000 and $60 000, depending upon patent complexity, and average somewhere around $15 000, this is not much disincentive. Second, despite the apparent fondness for lawyers to speak in a language the rest of us have difficulty understanding, patents should be written clearly and in a language all of us can understand. For example, I can imagine searching the patent database for the presence of the word ‘wheel’, but would never think to search for its legally equivalent phrase ‘circular transportation facilitation device’. Use Occam's razor liberally—if a simpler phrase can be used in place of a more complex one, force the inventors to use it and reject their patent if they do not. This will help reduce the vague patents and save litigation time. Third, because it is harder to search for previously implemented algorithms, the bar for awarding software patents should be raised. It is hard enough to compile prior art in non-software fields, but there is no searchable code archive and even if there were, there are many different programmatic ways of accomplishing the same end result in many different languages. Patentable software inventions should be substantially different from prior art and very non-obvious. If you task 10 programmers to solve the same general problem the invention claims to solve, and at least half come up with the same solution in a day, it is obvious. Fourth, perhaps the patenting system could borrow something from the academic community and begin peer-reviewing patents, requesting several independent evaluations from outside experts. One of the problems with the patent office is that their expertise and time is limited compared with the number of applications they receive. Peer-reviewers would be paid for their time, make their names known on the application and their roles be limited to providing advice to the examiner rather than making the final decision (to avoid conflicts of interest). After turning their comments in, they would also be able to see and comment on all the other reviewers. This way, if there are any particularly good or bad reviews, then this helps highlight them by further peer-reviewing the peer-reviewers. Finally, perhaps journals should enforce a difficult choice for most academics—to publish OR patent, but not both. A patent is a publication, after all, even if not peer-reviewed. We live in the digital age, and most of the new innovations coming out are computer related. The call to end software patents is growing, but doing so may be a Faustian bargain. How will we protect and incentivize the next great idea to become a reality—to move from paper to practice? Clearly, though, the list of problems associated with software patents is growing and the list of solutions needs to catch up. I would like to thank Steven Salzberg and John Quackenbush for a lively and valuable discussion on the merits, ethics and issues of software patents and Anil V. Gollahalli for his helpful legal review. Conflict of Interest: The author is listed as an inventor or co-inventor on three software-related patents, all of which are pending approval by the USPTO.

MOTIVATION: The use or study of chemical compounds permeates almost every scientific field and in each of them, the amount of textual information is growing rapidly. There is a need to accurately identify chemical names within text for a number of informatics efforts such as database curation, report summarization, tagging of named entities and keywords, or the development/curation of reference databases. RESULTS: A first-order Markov Model (MM) was evaluated for its ability to distinguish chemical names from words, yielding approximately 93% recall in recognizing chemical terms and approximately 99% precision in rejecting non-chemical terms on smaller test sets. However, because total false-positive events increase with the number of words analyzed, the scalability of name recognition was measured by processing 13.1 million MEDLINE records. The method yielded precision ranges from 54.7% to 100%, depending upon the cutoff score used, averaging 82.7% for approximately 1.05 million putative chemical terms extracted. Extracted chemical terms were analyzed to estimate the number of spelling variants per term, which correlated with the total number of times the chemical name appeared in MEDLINE. This variability in term construction was found to affect both information retrieval and term mapping when using PubMed and Ovid.

MOTIVATION: Short sequence patterns frequently define regions of biological interest (binding sites, immune epitopes, primers, etc.), yet a large fraction of this information exists only within the scientific literature and is thus difficult to locate via conventional means (e.g. keyword queries or manual searches). We describe herein a system to accurately identify and classify sequence patterns from within large corpora using an n-gram Markov model (MM). RESULTS: As expected, on test sets we found that identification of sequences with limited alphabets and/or regular structures such as nucleic acids (non-ambiguous) and peptide abbreviations (3-letter) was highly accurate, whereas classification of symbolic (1-letter) peptide strings with more complex alphabets was more problematic. The MM was used to analyze two very large, sequence-containing corpora: over 7.75 million Medline abstracts and 9000 full-text articles from Journal of Virology. Performance was benchmarked by comparing the results with Journal of Virology entries in two existing manually curated databases: VirOligo and the HLA Ligand Database. Performance estimates were 98 +/- 2% precision/84% recall for primer identification and classification and 67 +/- 6% precision/85% recall for peptide epitopes. We also find a dramatic difference between the amounts of sequence-related data reported in abstracts versus full text. Our results suggest that automated extraction and classification of sequence elements is a promising, low-cost means of sequence database curation and annotation. AVAILABILITY: MM routine and datasets are available upon request.

There is an enormous amount of information encoded in each genome--enough to create living, responsive and adaptive organisms. Raw sequence data alone is not enough to understand function, mechanisms or interactions. Changes in a single base pair can lead to disease, such as sickle-cell anemia, while some large megabase deletions have no apparent phenotypic effect. Genomic features are varied in their data types and annotation of these features is spread across multiple databases. Herein, we develop a method to automate exploration of genomes by iteratively exploring sequence data for correlations and building upon them. First, to integrate and compare different annotation sources, a sequence matrix (SM) is developed to contain position-dependant information. Second, a classification tree is developed for matrix row types, specifying how each data type is to be treated with respect to other data types for analysis purposes. Third, correlative analyses are developed to analyze features of each matrix row in terms of the other rows, guided by the classification tree as to which analyses are appropriate. A prototype was developed and successful in detecting coinciding genomic features among genes, exons, repetitive elements and CpG islands.

Abstract The MCBIOS 2004 conference brought together regional researchers and students in biology, computer science and bioinformatics on October 7th-9th 2004 to present their latest work. This editorial describes the conference itself and introduces the twelve peer-reviewed manuscripts accepted for publication in the Proceedings of the MCBIOS 2004 Conference. These manuscripts included new methods for analysis of high-throughput gene expression experiments, EST clustering, analysis of mass spectrometry data and genomic analysis

BACKGROUND: This paper describes a microarray study including data quality control, data analysis and the analysis of the mechanism of toxicity (MOT) induced by 1-methyl-4-phenylpyridinium (MPP+) in a rat adrenal pheochromocytoma cell line (PC12 cells) using bioinformatics tools. MPP+ depletes dopamine content and elicits cell death in PC12 cells. However, the mechanism of MPP+-induced neurotoxicity is still unclear. RESULTS: In this study, Agilent rat oligo 22K microarrays were used to examine alterations in gene expression of PC12 cells after 500 muM MPP+ treatment. Relative gene expression of control and treated cells represented by spot intensities on the array chips was analyzed using bioinformatics tools. Raw data from each array were input into the NCTR ArrayTrack database, and normalized using a Lowess normalization method. Data quality was monitored in ArrayTrack. The means of the averaged log ratio of the paired samples were used to identify the fold changes of gene expression in PC12 cells after MPP+ treatment. Our data showed that 106 genes and ESTs (Expressed Sequence Tags) were changed 2-fold and above with MPP+ treatment; among these, 75 genes had gene symbols and 59 genes had known functions according to the Agilent gene Refguide and ArrayTrack-linked gene library. The mechanism of MPP+-induced toxicity in PC12 cells was analyzed based on their genes functions, biological process, pathways and previous published literatures. CONCLUSION: Multiple pathways were suggested to be involved in the mechanism of MPP+-induced toxicity, including oxidative stress, DNA and protein damage, cell cycling arrest, and apoptosis.

Lisa M. Schilling, Jonathan D. Wren, Robert P. Dellavalle; Letter to the Editor, Bioinformatics, Volume 20, Issue 17, 22 November 2004, Pages 2903, https://doi.

MOTIVATION: The advent of the World Wide Web has enabled unprecedented supplementation of traditional journal publications, allowing access to resources, such as video, sound, software, databases, datasets too large to publish, and even supplementary information and discussion. However, unlike traditional publications, continued availability of these online resources is not guaranteed. An automated survey was conducted to quantify the growth in Uniform Resource Locators (URLs) published to date in MEDLINE abstracts, their current availability and distribution by journal. RESULTS: Of 1630 unique URLs identified, formatting and/or spelling errors were detected within 201 (12%) of them as published. After corrections were made, a survey revealed that approximately 63% of these URLs were consistently available, and another 19% were available intermittently. The rate of failure was far worse for anonymous login to FTP sites, with only 12 of 33 sites (36%) responding. This survey also shows that journals vary disproportionately in the number of web citations published, suggesting policy implementation among a few could have a profound impact overall. Out of the 306 journals with a URL published in an abstract, Bioinformatics published the most (12% of total). AVAILABILITY: URL database and program available by request.

MOTIVATION: New relationships are often implicit from existing information, but the amount and growth of published literature limits the scope of analysis an individual can accomplish. Our goal was to develop and test a computational method to identify relationships within scientific reports, such that large sets of relationships between unrelated items could be sought out and statistically ranked for their potential relevance as a set. RESULTS: We first construct a network of tentative relationships between 'objects' of biomedical research interest (e.g. genes, diseases, phenotypes, chemicals) by identifying their co-occurrences within all electronically available MEDLINE records. Relationships shared by two unrelated objects are then ranked against a random network model to estimate the statistical significance of any given grouping. When compared against known relationships, we find that this ranking correlates with both the probability and frequency of object co-occurrence, demonstrating the method is well suited to discover novel relationships based upon existing shared relationships. To test this, we identified compounds whose shared relationships predicted they might affect the development and/or progression of cardiac hypertrophy. When laboratory tests were performed in a rodent model, chlorpromazine was found to reduce the progression of cardiac hypertrophy.

MOTIVATION: There is a general scientific need to be able to identify and evaluate what any given set of 'objects' (e.g. genes, phenotypes, chemicals, diseases) has in common. Whether it is to classify, expand upon or identify commonalities and functional groupings, informational needs can be diverse and the best source to identify relationships among a potentially heterogeneous set of objects is the scientific literature. RESULTS: We first establish a network of related objects by their co-occurrence within MEDLINE records. A set of objects within this network can then be queried to identify shared relationships, and a method is presented to score their statistical relevance by comparing observed frequencies with what would be expected in a random network model. Using Gene Ontology (GO) categories, we demonstrate that this method enables a quantitative ranking of the 'cohesiveness' of a set of objects and, importantly, allows other objects related to this set to be identified and evaluated for their 'cohesion' to it. Supplemental information: A list of ranked genes related to each GO category analyzed can be found at http://innovation.swmed.edu/IRIDESCENT/GO_relationships.htm

BACKGROUND: Within the peer-reviewed literature, associations between two things are not always recognized until commonalities between them become apparent. These commonalities can provide justification for the inference of a new relationship where none was previously known, and are the basis of most observation-based hypothesis formation. It has been shown that the crux of the problem is not finding inferable associations, which are extraordinarily abundant given the scale-free networks that arise from literature-based associations, but determining which ones are informative. The Mutual Information Measure (MIM) is a well-established method to measure how informative an association is, but is limited to direct (i.e. observable) associations. RESULTS: Herein, we attempt to extend the calculation of mutual information to indirect (i.e. inferable) associations by using the MIM of shared associations. Objects of general research interest (e.g. genes, diseases, phenotypes, drugs, ontology categories) found within MEDLINE are used to create a network of associations for evaluation. CONCLUSIONS: Mutual information calculations can be effectively extended into implied relationships and a significance cutoff estimated from analysis of random word networks. Of the models tested, the shared minimum MIM (MMIM) model is found to correlate best with the observed strength and frequency of known associations. Using three test cases, the MMIM method tends to rank more specific relationships higher than counting the number of shared relationships within a network.

ARROGANT (ARRay OrGANizing Tool) is a software tool developed to facilitate the identification, annotation and comparison of large collections of genes or clones. The objective is to enable users to compile gene/clone collections from different databases, allowing them to design experiments and analyze the collections as well as associated experimental data efficiently. ARROGANT can relate different sequence identifiers to their common reference sequence using the UniGene database, allowing for the comparison of data from two different microarray experiments. ARROGANT has been successfully used to analyze microarray expression data for colon cancer, to compile genes potentially related to cardiac diseases for subsequent resequencing (to identify single nucleotide polymorphisms, SNPs), to design a new comprehensive human cDNA microarray for cancer, to combine and compare expression data generated by different microarrays and to provide annotation for genes on custom and Affymetrix chips.