MOTIVATION: Genetic analysis of cancer regularly includes two or more samples from the same patient. Somatic copy number alterations leading to allelic imbalance (AI) play a critical role in cancer initiation and progression. Directional analysis and visualization of the alleles in imbalance in multi-sample settings allow for inference of recurrent mutations, providing insights into mutation rates, clonality and the genomic architecture and etiology of cancer. RESULTS: The REpeat Chromosomal changes Uncovered by Reflection (RECUR) is an R application for the comparative analysis of AI profiles derived from SNP array and next-generation sequencing data. The algorithm accepts genotype calls and 'B allele' frequencies (BAFs) from at least two samples derived from the same individual. For a predefined set of genomic regions with AI, RECUR compares BAF values among samples. In the presence of AI, the expected value of a BAF can shift in two possible directions, reflecting an increased or decreased abundance of the maternal haplotype, relative to the paternal. The phenomenon of opposite haplotype shifts, or 'mirrored subclonal allelic imbalance', is a form of heterogeneity, and has been linked to clinico-pathological features of cancer. RECUR detects such genomic segments of opposite haplotypes in imbalance and plots BAF values for all samples, using a two-color scheme for intuitive visualization. AVAILABILITY AND IMPLEMENTATION: RECUR is available as an R application. Source code and documentation are available at scheet.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: 'Next-generation' (NGS) sequencing has wide application in medical genetics, including the detection of somatic variation in cancer. The Ion Torrent-based (IONT) platform is among NGS technologies employed in clinical, research and diagnostic settings. However, identifying mutations from IONT deep sequencing with high confidence has remained a challenge. We compared various computational variant-calling methods to derive a variant identification pipeline that may improve the molecular diagnostic and research utility of IONT. RESULTS: Using IONT, we surveyed variants from the 409-gene Comprehensive Cancer Panel in whole-section tumors, intra-tumoral biopsies and matched normal samples obtained from frozen tissues and blood from four early-stage non-small cell lung cancer (NSCLC) patients. We used MuTect, Varscan2, IONT's proprietary Ion Reporter, and a simple subtraction we called "Poor Man's Caller." Together these produced calls at 637 loci across all samples. Visual validation of 434 called variants was performed, and performance of the methods assessed individually and in combination. Of the subset of inspected putative variant calls (n=223) in genomic regions that were not intronic or intergenic, 68 variants (30%) were deemed valid after visual inspection. Among the individual methods, the Ion Reporter method offered perhaps the most reasonable tradeoffs. Ion Reporter captured 83% of all discovered variants; 50% of its variants were visually validated. Aggregating results from multiple packages offered varied improvements in performance. CONCLUSIONS: Overall, Ion Reporter offered the most attractive performance among the individual callers. This study suggests combined strategies to maximize sensitivity and positive predictive value in variant calling using IONT deep sequencing.

MOTIVATION: The detection of subtle genomic allelic imbalance events has many potential applications. For example, identifying cancer-associated allelic imbalanced regions in low tumor-cellularity samples or in low-proportion tumor subclones can be used for early cancer detection, prognostic assessment and therapeutic selection in cancer patients. We developed hapLOHseq for the detection of subtle allelic imbalance events from next-generation sequencing data. RESULTS: Our method identified events of 10 megabases or greater occurring in as little as 16% of the sample in exome sequencing data (at 80×) and 4% in whole genome sequencing data (at 30×), far exceeding the capabilities of existing software. We also found hapLOHseq to be superior at detecting large chromosomal changes across a series of pancreatic samples from TCGA. AVAILABILITY AND IMPLEMENTATION: hapLOHseq is available at scheet.org/software, distributed under an open source MIT license. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: Many variable selection techniques have been proposed for the clustering of gene expression data. While these methods tend to filter out irrelevant genes and identify informative genes that contribute to a clustering solution, they are based on criteria that do not consider the potential interactive influence among individual genes. Motivated by ensemble clustering, there is a strong interest in leveraging the structure of gene networks for gene selection, so that the relationship information between genes can be effectively utilized, while the selected genes are expected to preserve all the possible clustering structures in the data. RESULTS: We present a new filter method that uses the gene connectivity in the gene co-expression network as the evaluation criteria for variable selection. The gene connectivity measures the importance of the genes in term of their expression similarity with others in the co-expression network. The hard threshold and soft threshold transformations are employed to construct the gene co-expression networks. Both simulation studies and real data analysis have shown that the network based on soft thresholding is more effective in selecting relevant variables and provides better clustering results compared to the hard thresholding transformation and two other canonical filter methods for variable selection. Furthermore, a new module analysis approach is proposed to reveal the higher order organization of the gene space, where the genes of a module share significant topological similarity and are associated with a consensus partition of the sample space. We demonstrate that the identified modules can lead to biologically meaningful sample partitions that might be missed by other methods. CONCLUSIONS: By leveraging the structure of gene co-expression network, first we propose a variable selection method that selects individual genes with top connectivity. Both simulation studies and real data application have demonstrated that our method has better performance in terms of the reliability of the selected genes and sample clustering results. In addition, we propose a module recovery method that can help discover novel sample partitions that might be hidden when performing clustering analyses using all available genes. The source code of our program is available at http://nba.uth.tmc.edu/homepage/liu/netVar/.

BACKGROUND: Currently available microRNA (miRNA) target prediction algorithms require the presence of a conserved seed match to the 5' end of the miRNA and limit the target sites to the 3' untranslated regions of mRNAs. However, it has been noted that these requirements may be too stringent, leading to a substantial number of missing targets. RESULTS: We have developed TargetS, a novel computational approach for predicting miRNA targets with the target sites located along entire gene sequences, which permits finding additional targets that are not located in the 3' un-translated regions. Our model is based on both canonical seed matching and non-canonical seed pairing, which discovers targets that allow one bit GU wobble. It does not rely on evolutionary conservation, so it allows the detection of species-specific miRNA-mRNA interactions and makes it suitable for analyzing un-conserved gene sequences. To test the performance of our approach, we have imported the widely used benchmark dataset revealing fold-changes in protein production corresponding to each of the five selected microRNAs. Compared to well-known miRNA target prediction tools, including TargetScanS, PicTar and MicroT_CDS, our method yields the highest sensitivity, while achieving a comparable level of accuracy. Human miRNA target predictions using our computational approach are available online at http://liubioinfolab.org/targetS/mirna.html CONCLUSIONS: A simple but powerful computational miRNA target prediction method is developed that is solely based on canonical and non-canonical seed matches without requiring evolutionary conservation of the target sites. Our method also expands the target search space to different gene regions, rather than limiting to 3'UTR only. This improves the sensitivity of miRNA target identification, while achieving a comparable accuracy with existing methods.

MOTIVATION: A major goal in genomic research is to identify genes that may jointly influence a biological response. From many years of intensive biomedical research, a large body of biological knowledge, or pathway information, has accumulated in available databases. There is a strong interest in leveraging these pathways to improve the statistical power and interpretability in studying gene networks associated with complex phenotypes. This prior information is a valuable complement to large-scale genomic data such as gene expression data generated from microarrays. However, it is a non-trivial task to effectively integrate available biological knowledge into gene expression data when reconstructing gene networks. RESULTS: In this article, we developed and applied a Lasso method from a Bayesian perspective, a method we call prior Lasso (pLasso), for the reconstruction of gene networks. In this method, we partition edges between genes into two subsets: one subset of edges is present in known pathways, whereas the other has no prior information associated. Our method assigns different prior distributions to each subset according to a modified Bayesian information criterion that incorporates prior knowledge on both the network structure and the pathway information. Simulation studies have indicated that the method is more effective in recovering the underlying network than a traditional Lasso method that does not use the prior information. We applied pLasso to microarray gene expression datasets, where we used information from the Pathway Commons (PC) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) as prior information for the network reconstruction, and successfully identified network hub genes associated with clinical outcome in cancer patients. AVAILABILITY: The source code is available at http://nba.uth.tmc.edu/homepage/liu/pLasso.

MOTIVATION: Storing, annotating and analyzing variants from next-generation sequencing projects can be difficult due to the availability of a wide array of data formats, tools and annotation sources, as well as the sheer size of the data files. Useful tools, including the GATK, ANNOVAR and BEDTools can be integrated into custom pipelines for annotating and analyzing sequence variants. However, building flexible pipelines that support the tracking of variants alongside their samples, while enabling updated annotation and reanalyses, is not a simple task. RESULTS: We have developed variant tools, a flexible annotation and analysis toolset that greatly simplifies the storage, annotation and filtering of variants and the analysis of the underlying samples. variant tools can be used to manage and analyze genetic variants obtained from sequence alignments, and the command-line driven toolset could be used as a foundation for building more sophisticated analytical methods. AVAILABILITY AND IMPLEMENTATION: variant tools consists of two command-line driven programs vtools and vtools_report. It is freely available at http://varianttools.sourceforge.net, distributed under a GPL license. CONTACT: [email protected].