SUMMARY: Complex genomic analyses often use sequences of simple set operations like intersection, overlap and nearest on genomic intervals. These operations, coupled with some custom programming, allow a wide range of analyses to be performed. To this end, we have written PyRanges, a data structure for representing and manipulating genomic intervals and their associated data in Python. Run single threaded on binary set operations, PyRanges is in median 2.3-9.6 times faster than the popular R GenomicRanges library and is equally memory efficient; run multi-threaded on 8 cores, our library is up to 123 times faster. PyRanges is therefore ideally suited both for individual analyses and as a foundation for future genomic libraries in Python. AVAILABILITY AND IMPLEMENTATION: PyRanges is available as open source under the MIT license at https://github.com/biocore-NTNU/pyranges and the documentation exists at https://biocore-NTNU.github.io/pyranges/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SUMMARY: Data from chromatin immunoprecipitation (ChIP) followed by high-throughput sequencing (ChIP-seq) generally contain either narrow peaks or broad and diffusely enriched domains. The SICER ChIP-seq caller has proven adept at finding diffuse domains in ChIP-seq data, but it is slow, requires much memory, needs manual installation steps and is hard to use. epic2 is a complete rewrite of SICER that is focused on speed, low memory overhead and ease-of-use. AVAILABILITY AND IMPLEMENTATION: The MIT-licensed code is available at https://github.com/biocore-ntnu/epic2. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Circular RNAs (circRNAs) are an abundant class of highly stable RNAs that can affect gene regulation by binding and preventing microRNAs (miRNAs) from regulating their messenger RNA (mRNA) targets. Mammals have thousands of circRNAs with predicted miRNA binding sites, but only two circRNAs have been verified as being actual miRNA sponges. As it is unclear whether these thousands of predicted miRNA binding sites are functional, we investigated whether miRNA seed sites within human circRNAs are under selective pressure. RESULTS: Using SNP data from the 1000 Genomes Project, we found a significant decrease in SNP density at miRNA seed sites compared with flanking sequences and random sites. This decrease was similar to that of miRNA seed sites in 3' untranslated regions, suggesting that many of the predicted miRNA binding sites in circRNAs are functional and under similar selective pressure as miRNA binding sites in mRNAs.

Alternative polyadenylation (APA) can for example occur when a protein-coding gene has several polyadenylation (polyA) signals in its last exon, resulting in messenger RNAs (mRNAs) with different 3' untranslated region (UTR) lengths. Different 3'UTR lengths can give different microRNA (miRNA) regulation such that shortened transcripts have increased expression. The APA process is part of human cells' natural regulatory processes, but APA also seems to play an important role in many human diseases. Although altered APA in disease can have many causes, we reasoned that mutations in DNA elements that are important for the polyA process, such as the polyA signal and the downstream GU-rich region, can be one important mechanism. To test this hypothesis, we identified single nucleotide polymorphisms (SNPs) that can create or disrupt APA signals (APA-SNPs). By using a data-integrative approach, we show that APA-SNPs can affect 3'UTR length, miRNA regulation, and mRNA expression--both between homozygote individuals and within heterozygote individuals. Furthermore, we show that a significant fraction of the alleles that cause APA are strongly and positively linked with alleles found by genome-wide studies to be associated with disease. Our results confirm that APA-SNPs can give altered gene regulation and that APA alleles that give shortened transcripts and increased gene expression can be important hereditary causes for disease.

BACKGROUND: Despite experiments showing that the number of microRNA (miRNA) target sites is critical for miRNA targeting, most existing methods focus on identifying individual miRNA target sites and do not model contributions of multiple target sites to miRNA regulation. To address this possible fault, we developed a miRNA target prediction model that recognizes the individual characteristics of functional binding sites and the global characteristics of miRNA-targeted mRNAs. RESULTS: Benchmark experiments showed that this two-step model generally had a higher overall performance than other established miRNA target prediction algorithms and that the model was especially suited to identify true miRNA targets among genes that all contain conserved target sites. CONCLUSIONS: This improved performance could partly be explained by the model not relying on conservation when predicting targets. The critical factors for the model's performance, however, were mRNA-level features that characterized the number and strength of individual target sites within the mRNA. The model is available for online predictions or as pre-computed predictions on the human genome http://tare.medisin.ntnu.no/mirna_target.

BACKGROUND: Gene expression measurements from breast cancer (BrCa) tumors are established clinical predictive tools to identify tumor subtypes, identify patients showing poor/good prognosis, and identify patients likely to have disease recurrence. However, diverse breast cancer datasets in conjunction with diagnostic clinical arrays show little overlap in the sets of genes identified. One approach to identify a set of consistently dysregulated candidate genes in these tumors is to employ meta-analysis of multiple independent microarray datasets. This allows one to compare expression data from a diverse collection of breast tumor array datasets generated on either cDNA or oligonucleotide arrays. RESULTS: We gathered expression data from 9 published microarray studies examining estrogen receptor positive (ER+) and estrogen receptor negative (ER-) BrCa tumor cases from the Oncomine database. We performed a meta-analysis and identified genes that were universally up or down regulated with respect to ER+ versus ER- tumor status. We surveyed both the proximal promoter and 3' untranslated regions (3'UTR) of our top-ranking genes in each expression group to test whether common sequence elements may contribute to the observed expression patterns. Utilizing a combination of known transcription factor binding sites (TFBS), evolutionarily conserved mammalian promoter and 3'UTR motifs, and microRNA (miRNA) seed sequences, we identified numerous motifs that were disproportionately represented between the two gene classes suggesting a common regulatory network for the observed gene expression patterns. CONCLUSION: Some of the genes we identified distinguish key transcripts previously seen in array studies, while others are newly defined. Many of the genes identified as overexpressed in ER- tumors were previously identified as expression markers for neoplastic transformation in multiple human cancers. Moreover, our motif analysis identified a collection of specific cis-acting target sites which may collectively play a role in the differential gene expression patterns observed in ER+ versus ER- breast cancer tumors. Importantly, the gene sets and associated DNA motifs provide a starting point with which to explore the mechanistic basis for the observed expression patterns in breast tumors.

MOTIVATION: Mature microRNAs (miRNAs) are processed from long hairpin transcripts. Even though it is only the first of several steps, the initial Drosha processing defines the mature product and is characteristic for all miRNA genes. Methods that can separate between true and false processing sites are therefore essential to miRNA gene discovery. RESULTS: We present a classifier that predicts 5' Drosha processing sites in hairpins that are candidate miRNAs. The classifier, called Microprocessor SVM, correctly predicts the processing site for 50% of known human 5' miRNAs, and 90% of its predictions are within two nucleotides of the true site. Another classifier that is trained on the output from the Microprocessor SVM outperforms existing methods for prediction of unconserved miRNAs. Reanalysis of characteristics and supporting evidence for a set of newly annotated miRNAs shows that some miRNAs may be misannotated. This suggests that expressed hairpins should not be annotated as miRNAs until they are verified to be Drosha and Dicer substrates. AVAILABILITY: The classifiers are publicly available at https://demo1.interagon.com/miRNA/

BACKGROUND: Protein remote homology detection is a central problem in computational biology. Most recent methods train support vector machines to discriminate between related and unrelated sequences and these studies have introduced several types of kernels. One successful approach is to base a kernel on shared occurrences of discrete sequence motifs. Still, many protein sequences fail to be classified correctly for a lack of a suitable set of motifs for these sequences. RESULTS: We introduce the GPkernel, which is a motif kernel based on discrete sequence motifs where the motifs are evolved using genetic programming. All proteins can be grouped according to evolutionary relations and structure, and the method uses this inherent structure to create groups of motifs that discriminate between different families of evolutionary origin. When tested on two SCOP benchmarks, the superfamily and fold recognition problems, the GPkernel gives significantly better results compared to related methods of remote homology detection. CONCLUSION: The GPkernel gives particularly good results on the more difficult fold recognition problem compared to the other methods. This is mainly because the method creates motif sets that describe similarities among subgroups of both the related and unrelated proteins. This rich set of motifs give a better description of the similarities and differences between different folds than do previous motif-based methods.

Multiple instruction stream-single data stream (MISD) architectures have not found many practical applications in supercomputing. We present a multiple instruction stream-multiple data stream (MIMD) cluster implementation that uses MISD search processors with extreme pattern mining performance as a building block. We use PCI cards that hold sixteen search processors with local memory to build a relatively small cluster of five PC nodes with six PCI cards each. This cluster can handle anything between 64 independent queries at 48 GB per second, to 30,720 independent queries at 100 MB per second. The cluster's performance characteristics are such that we can easily scale the system to 1015operations per second with containable overhead using just 100 nodes. Further, the solution has lower power consumption and memory bandwidth requirements than comparable technologies.

MOTIVATION: Both small interfering RNAs (siRNAs) and antisense oligonucleotides can selectively block gene expression. Although the two methods rely on different cellular mechanisms, these methods share the common property that not all oligonucleotides (oligos) are equally effective. That is, if mRNA target sites are picked at random, many of the antisense or siRNA oligos will not be effective. Algorithms that can reliably predict the efficacy of candidate oligos can greatly reduce the cost of knockdown experiments, but previous attempts to predict the efficacy of antisense oligos have had limited success. Machine learning has not previously been used to predict siRNA efficacy. RESULTS: We develop a genetic programming based prediction system that shows promising results on both antisense and siRNA efficacy prediction. We train and evaluate our system on a previously published database of antisense efficacies and our own database of siRNA efficacies collected from the literature. The best models gave an overall correlation between predicted and observed efficacy of 0.46 on both antisense and siRNA data. As a comparison, the best correlations of support vector machine classifiers trained on the same data were 0.40 and 0.30, respectively.

Many applications require searching for multiple patterns in large data streams for which there is no preprocessed index to rely on for efficient lookups. An multiple instruction stream-single data stream (MISD) VLSI architecture that is based on a recursive divide and conquer approach to pattern matching is proposed. This architecture allows searching for multiple patterns simultaneously. The patterns can be constructed much like regular expressions, and add features such as requiring subpatterns to match in a specific order with some fuzzy distance between them, and the ability to allow errors according to prescribed thresholds, or ranges of such. The current implementation permits up to 127 simultaneous patterns at a clock frequency of 100 MHz, and does 1.024/spl times/10/sup 11/ character comparisons per second.