MOTIVATION: Convolutional neural networks (CNN) have outperformed conventional methods in modeling the sequence specificity of DNA-protein binding. Yet inappropriate CNN architectures can yield poorer performance than simpler models. Thus an in-depth understanding of how to match CNN architecture to a given task is needed to fully harness the power of CNNs for computational biology applications. RESULTS: We present a systematic exploration of CNN architectures for predicting DNA sequence binding using a large compendium of transcription factor datasets. We identify the best-performing architectures by varying CNN width, depth and pooling designs. We find that adding convolutional kernels to a network is important for motif-based tasks. We show the benefits of CNNs in learning rich higher-order sequence features, such as secondary motifs and local sequence context, by comparing network performance on multiple modeling tasks ranging in difficulty. We also demonstrate how careful construction of sequence benchmark datasets, using approaches that control potentially confounding effects like positional or motif strength bias, is critical in making fair comparisons between competing methods. We explore how to establish the sufficiency of training data for these learning tasks, and we have created a flexible cloud-based framework that permits the rapid exploration of alternative neural network architectures for problems in computational biology. AVAILABILITY AND IMPLEMENTATION: All the models analyzed are available at http://cnn.csail.mit.edu CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

We show that existing RNA-seq, DNase-seq, and ChIP-seq data exhibit overdispersed per-base read count distributions that are not matched to existing computational method assumptions. To compensate for this overdispersion we introduce a nonparametric and universal method for processing per-base sequencing read count data called FIXSEQ. We demonstrate that FIXSEQ substantially improves the performance of existing RNA-seq, DNase-seq, and ChIP-seq analysis tools when compared with existing alternatives.

Regulatory proteins can bind to different sets of genomic targets in various cell types or conditions. To reliably characterize such condition-specific regulatory binding we introduce MultiGPS, an integrated machine learning approach for the analysis of multiple related ChIP-seq experiments. MultiGPS is based on a generalized Expectation Maximization framework that shares information across multiple experiments for binding event discovery. We demonstrate that our framework enables the simultaneous modeling of sparse condition-specific binding changes, sequence dependence, and replicate-specific noise sources. MultiGPS encourages consistency in reported binding event locations across multiple-condition ChIP-seq datasets and provides accurate estimation of ChIP enrichment levels at each event. MultiGPS's multi-experiment modeling approach thus provides a reliable platform for detecting differential binding enrichment across experimental conditions. We demonstrate the advantages of MultiGPS with an analysis of Cdx2 binding in three distinct developmental contexts. By accurately characterizing condition-specific Cdx2 binding, MultiGPS enables novel insight into the mechanistic basis of Cdx2 site selectivity. Specifically, the condition-specific Cdx2 sites characterized by MultiGPS are highly associated with pre-existing genomic context, suggesting that such sites are pre-determined by cell-specific regulatory architecture. However, MultiGPS-defined condition-independent sites are not predicted by pre-existing regulatory signals, suggesting that Cdx2 can bind to a subset of locations regardless of genomic environment. A summary of this paper appears in the proceedings of the RECOMB 2014 conference, April 2-5.

BACKGROUND: Modern genetics has been transformed by high-throughput sequencing. New experimental designs in model organisms involve analyzing many individuals, pooled and sequenced in groups for increased efficiency. However, the uncertainty from pooling and the challenge of noisy sequencing data demand advanced computational methods. RESULTS: We present MULTIPOOL, a computational method for genetic mapping in model organism crosses that are analyzed by pooled genotyping. Unlike other methods for the analysis of pooled sequence data, we simultaneously consider information from all linked chromosomal markers when estimating the location of a causal variant. Our use of informative sequencing reads is formulated as a discrete dynamic Bayesian network, which we extend with a continuous approximation that allows for rapid inference without a dependence on the pool size. MULTIPOOL generalizes to include biological replicates and case-only or case-control designs for binary and quantitative traits. CONCLUSIONS: Our increased information sharing and principled inclusion of relevant error sources improve resolution and accuracy when compared to existing methods, localizing associations to single genes in several cases. MULTIPOOL is freely available at http://cgs.csail.mit.edu/multipool/.