Mouse Transcription Factor Research Articles

Identification of mammalian transcription factors that bind to inaccessible chromatin.

Transcription factors (TFs) are proteins that affect gene expression by binding to regulatory regions of DNA in a sequence specific manner. The binding of TFs to DNA is controlled by many factors, including the DNA sequence, concentration of TF, chromatin accessibility and co-factors. Here, we systematically investigated the binding mechanism of hundreds of TFs by analysing ChIP-seq data with our explainable statistical model, ChIPanalyser. This tool uses as inputs the DNA sequence binding motif; the capacity to distinguish between strong and weak binding sites; the concentration of TF; and chromatin accessibility. We found that approximately one third of TFs are predicted to bind the genome in a DNA accessibility independent fashion, which includes TFs that can open the chromatin, their co-factors and TFs with similar motifs. Our model predicted this to be the case when the TF binds to its strongest binding regions in the genome, and only a small number of TFs have the capacity to bind dense chromatin at their weakest binding regions, such as CTCF, USF2 and CEBPB. Our study demonstrated that the binding of hundreds of human and mouse TFs is predicted by ChIPanalyser with high accuracy and showed that many TFs can bind dense chromatin.

GNet: An integrated context-aware neural framework for transcription factor binding signal at single nucleotide resolution prediction.

Transcription factors (TFs) are important factors that regulate gene expression. Revealing the mechanism affecting the binding specificity of TFs is the key to understanding gene regulation. Most of the previous studies focus on TF-DNA binding sites at the sequence level, and they seldom utilize the contextual features of DNA sequences. In this paper, we develop an integrated spatiotemporal context-aware neural network framework, named GNet, for predicting TF-DNA binding signal at single nucleotide resolution by achieving three tasks: single nucleotide resolution signal prediction, identification of binding regions at the sequence level, and TF-DNA binding motif prediction. GNet extracts implicit spatial contextual information with a gated highway neural mechanism, which captures large context multi-level patterns using linear shortcut connections, and the idea of it permeates the encoder and decoder parts of GNet. The improved dual external attention mechanism, which learns implicit relationships both within and among samples, and improves the performance of the model. Experimental results on 53 human TF ChIP-seq datasets and 6 chromatin accessibility ATAC-seq datasets shows that GNet outperforms the state-of-the-art methods in the three tasks, and the results of cross-species studies on 15 human and 18 mouse TF datasets of the corresponding TF families indicate that GNet also shows the best performance in cross-species prediction over the competitive methods.

Survey of the Transcription Factor Responses of Mouse Lung Alveolar Macrophages to Pneumocystis murina.

Pneumocystis jirovecii is a fungal pathogen that can cause life-threatening infections in individuals who are immunocompromised. Acquired via inhalation, upon entering the respiratory tract, the fungi first encounter innate immune cells such as alveolar macrophages (AMs). Relatively little is known about the AM cellular responses to the organism, and particularly transcription factor (TF) profiles leading to early host responses during infection. Utilizing the Mouse Transcription Factors RT2 Profiler™ PCR Array, we report an initial TF survey of these macrophage and Pneumocystis interactions. Expression levels of a panel of mouse TFs were compared between unstimulated and Pneumocystis murina-stimulated AMs. Interestingly, a number of TFs previously implicated in pathogen–host cell interactions were highly up- or downregulated, including hif1a and Pparg. qPCR experiments were further conducted to verify the results of these surveyed transcripts. Furthermore, with immunoblotting, we show that HIF-1A and PPAR-γ are indeed significantly upregulated and downregulated, respectively. Lastly, and importantly, we report that in the mouse model of Pneumocystis pneumonia (PCP), which mimics human Pneumocystis jirovecii pneumonia (PJP), qPCR analysis of Pneumocystis murina lungs also mimic the initial TF profile analysis, suggesting an importance for these TFs in immunocompromised hosts with Pneumocystis pneumonia. These data demonstrate the use of TF profiling in host AMs and Pneumocystis organism interactions that may lead to a better understanding of the specific inflammatory responses of the host to Pneumocystis pneumonia and may inform novel strategies for potential therapeutics.

TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions.

Transcription factors (TFs) are major trans-acting factors in transcriptional regulation. Therefore, elucidating TF–target interactions is a key step toward understanding the regulatory circuitry underlying complex traits such as human diseases. We previously published a reference TF–target interaction database for humans—TRRUST (Transcriptional Regulatory Relationships Unraveled by Sentence-based Text mining)—which was constructed using sentence-based text mining, followed by manual curation. Here, we present TRRUST v2 (www.grnpedia.org/trrust) with a significant improvement from the previous version, including a significantly increased size of the database consisting of 8444 regulatory interactions for 800 TFs in humans. More importantly, TRRUST v2 also contains a database for TF–target interactions in mice, including 6552 TF–target interactions for 828 mouse TFs. TRRUST v2 is also substantially more comprehensive and less biased than other TF–target interaction databases. We also improved the web interface, which now enables prioritization of key TFs for a physiological condition depicted by a set of user-input transcriptional responsive genes. With the significant expansion in the database size and inclusion of the new web tool for TF prioritization, we believe that TRRUST v2 will be a versatile database for the study of the transcriptional regulation involved in human diseases.

SMiLE-seq identifies binding motifs of single and dimeric transcription factors.

Resolving the DNA-binding specificities of transcription factors (TFs) is of critical value for understanding gene regulation. Here, we present a novel, semiautomated protein-DNA interaction characterization technology, selective microfluidics-based ligand enrichment followed by sequencing (SMiLE-seq). SMiLE-seq is neither limited by DNA bait length nor biased toward strong affinity binders; it probes the DNA-binding properties of TFs over a wide affinity range in a fast and cost-effective fashion. We validated SMiLE-seq by analyzing 58 full-length human, mouse, and Drosophila TFs from distinct structural classes. All tested TFs yielded DNA-binding models with predictive power comparable to or greater than that of other in vitro assays. De novo motif discovery on all JUN-FOS heterodimers and several nuclear receptor-TF complexes provided novel insights into partner-specific heterodimer DNA-binding preferences. We also successfully analyzed the DNA-binding properties of uncharacterized human C2H2 zinc-finger proteins and validated several using ChIP-exo.

CLIP-GENE: a web service of the condition specific context-laid integrative analysis for gene prioritization in mouse TF knockout experiments.

MotivationTranscriptome data from the gene knockout experiment in mouse is widely used to investigate functions of genes and relationship to phenotypes. When a gene is knocked out, it is important to identify which genes are affected by the knockout gene. Existing methods, including differentially expressed gene (DEG) methods, can be used for the analysis. However, existing methods require cutoff values to select candidate genes, which can produce either too many false positives or false negatives. This hurdle can be addressed either by improving the accuracy of gene selection or by providing a method to rank candidate genes effectively, or both. Prioritization of candidate genes should consider the goals or context of the knockout experiment. As of now, there are no tools designed for both selecting and prioritizing genes from the mouse knockout data. Hence, the necessity of a new tool arises.ResultsIn this study, we present CLIP-GENE, a web service that selects gene markers by utilizing differentially expressed genes, mouse transcription factor (TF) network, and single nucleotide variant information. Then, protein-protein interaction network and literature information are utilized to find genes that are relevant to the phenotypic differences. One of the novel features is to allow researchers to specify their contexts or hypotheses in a set of keywords to rank genes according to the contexts that the user specify. We believe that CLIP-GENE will be useful in characterizing functions of TFs in mouse experiments.Availability http://epigenomics.snu.ac.kr/CLIP-GENE ReviewersThis article was reviewed by Dr. Lee and Dr. Pongor.Electronic supplementary materialThe online version of this article (doi:10.1186/s13062-016-0158-x) contains supplementary material, which is available to authorized users.

Generation and gene expression profiling of 48 transcription-factor-inducible mouse embryonic stem cell lines.

Mouse embryonic stem cells (ESCs) can differentiate into a wide range – and possibly all cell types in vitro, and thus provide an ideal platform to study systematically the action of transcription factors (TFs) in cell differentiation. Previously, we have generated and analyzed 137 TF-inducible mouse ESC lines. As an extension of this “NIA Mouse ESC Bank,” we generated and characterized 48 additional mouse ESC lines, in which single TFs in each line could be induced in a doxycycline-controllable manner. Together, with the previous ESC lines, the bank now comprises 185 TF-manipulable ESC lines (>10% of all mouse TFs). Global gene expression (transcriptome) profiling revealed that the induction of individual TFs in mouse ESCs for 48 hours shifts their transcriptomes toward specific differentiation fates (e.g., neural lineages by Myt1 Isl1, and St18; mesodermal lineages by Pitx1, Pitx2, Barhl2, and Lmx1a; white blood cells by Myb, Etv2, and Tbx6, and ovary by Pitx1, Pitx2, and Dmrtc2). These data also provide and lists of inferred target genes of each TF and possible functions of these TFs. The results demonstrate the utility of mouse ESC lines and their transcriptome data for understanding the mechanism of cell differentiation and the function of TFs.

Transcription Factors Expressed in Mouse Cochlear Inner and Outer Hair Cells.

Regulation of gene expression is essential to determining the functional complexity and morphological diversity seen among different cells. Transcriptional regulation is a crucial step in gene expression regulation because the genetic information is directly read from DNA by sequence-specific transcription factors (TFs). Although several mouse TF databases created from genome sequences and transcriptomes are available, a cell type-specific TF database from any normal cell populations is still lacking. We identify cell type-specific TF genes expressed in cochlear inner hair cells (IHCs) and outer hair cells (OHCs) using hair cell-specific transcriptomes from adult mice. IHCs and OHCs are the two types of sensory receptor cells in the mammalian cochlea. We show that 1,563 and 1,616 TF genes are respectively expressed in IHCs and OHCs among 2,230 putative mouse TF genes. While 1,536 are commonly expressed in both populations, 73 genes are differentially expressed (with at least a twofold difference) in IHCs and 13 are differentially expressed in OHCs. Our datasets represent the first cell type-specific TF databases for two populations of sensory receptor cells and are key informational resources for understanding the molecular mechanism underlying the biological properties and phenotypical differences of these cells.

Linc-YY1 promotes myogenic differentiation and muscle regeneration through an interaction with the transcription factor YY1.

Little is known how lincRNAs are involved in skeletal myogenesis. Here we describe the discovery of Linc-YY1 from the promoter of the transcription factor (TF) Yin Yang 1 (YY1) gene. We demonstrate that Linc-YY1 is dynamically regulated during myogenesis in vitro and in vivo. Gain or loss of function of Linc-YY1 in C2C12 myoblasts or muscle satellite cells alters myogenic differentiation and in injured muscles has an impact on the course of regeneration. Linc-YY1 interacts with YY1 through its middle domain, to evict YY1/Polycomb repressive complex (PRC2) from target promoters, thus activating the gene expression in trans. In addition, Linc-YY1 also regulates PRC2-independent function of YY1. Finally, we identify a human Linc-YY1 orthologue with conserved function and show that many human and mouse TF genes are associated with lincRNAs that may modulate their activity. Altogether, we show that Linc-YY1 regulates skeletal myogenesis and uncover a previously unappreciated mechanism of gene regulation by lincRNA.

HOCOMOCO: expansion and enhancement of the collection of transcription factor binding sites models.

Models of transcription factor (TF) binding sites provide a basis for a wide spectrum of studies in regulatory genomics, from reconstruction of regulatory networks to functional annotation of transcripts and sequence variants. While TFs may recognize different sequence patterns in different conditions, it is pragmatic to have a single generic model for each particular TF as a baseline for practical applications. Here we present the expanded and enhanced version of HOCOMOCO (http://hocomoco.autosome.ru and http://www.cbrc.kaust.edu.sa/hocomoco10), the collection of models of DNA patterns, recognized by transcription factors. HOCOMOCO now provides position weight matrix (PWM) models for binding sites of 601 human TFs and, in addition, PWMs for 396 mouse TFs. Furthermore, we introduce the largest up to date collection of dinucleotide PWM models for 86 (52) human (mouse) TFs. The update is based on the analysis of massive ChIP-Seq and HT-SELEX datasets, with the validation of the resulting models on in vivo data. To facilitate a practical application, all HOCOMOCO models are linked to gene and protein databases (Entrez Gene, HGNC, UniProt) and accompanied by precomputed score thresholds. Finally, we provide command-line tools for PWM and diPWM threshold estimation and motif finding in nucleotide sequences.

Conservation of trans-acting circuitry during mammalian regulatory evolution.

The basic body plan and major physiological axes have been highly conserved during mammalian evolution, yet only a small fraction of the human genome sequence appears to be subject to evolutionary constraint. To quantify cis- versus trans-acting contributions to mammalian regulatory evolution, we performed genomic DNase I footprinting of the mouse genome across 25 cell and tissue types, collectively defining ∼8.6 million transcription factor (TF) occupancy sites at nucleotide resolution. Here we show that mouse TF footprints conjointly encode a regulatory lexicon that is ∼95% similar with that derived from human TF footprints. However, only ∼20% of mouse TF footprints have human orthologues. Despite substantial turnover of the cis-regulatory landscape, nearly half of all pairwise regulatory interactions connecting mouse TF genes have been maintained in orthologous human cell types through evolutionary innovation of TF recognition sequences. Furthermore, the higher-level organization of mouse TF-to-TF connections into cellular network architectures is nearly identical with human. Our results indicate that evolutionary selection on mammalian gene regulation is targeted chiefly at the level of trans-regulatory circuitry, enabling and potentiating cis-regulatory plasticity.Supplementary informationThe online version of this article (doi:10.1038/nature13972) contains supplementary material, which is available to authorized users.

Genomics: mice in the ENCODE spotlight.

Following on from affiliated projects in humans and model invertebrates, the Mouse ENCODE Project presents comprehensive data sets on genome regulation in this key mammalian model. See Articles p.355, p.365, p.371 & Letter p.402 The mouse is the premier model organism in biomedical research. To gain greater insights into the shared and species-specific transcriptional and cellular regulatory programs, the Mouse ENCODE Consortium has mapped transcription, DNase I hypersensitivity, transcription factor binding, chromatin modifications and replication domains throughout the mouse genome in diverse cell and tissue types. These finding are compared with the corresponding human data to confirm substantial conservation in the newly annotated potential functional sequences, and to reveal pronounced divergence of other sequences involved in transcriptional regulation, chromatin state and higher order chromatin organization. The data and their analyses provide a valuable resource for research into mammalian biology and mechanisms of human diseases. Having generated genomic DNase I footprinting data of the mouse genome across 25 cell and tissue types, these authors use these data to quantify cis-versus-trans regulatory contributions to mammalian regulatory evolution. They describe more than 600 motifs that collectively are over 95% similar to that recognized in vivo by human transcription factors (TFs). Despite substantial turnover of the cis-regulatory landscape around each TF gene, nearly half of all pairwise regulatory interactions connecting mouse TF genes have been maintained in orthologous human cell types through evolutionary innovation of TF recognition sequences. Conservation between mouse and human TF regulatory networks is particularly similar at the highest organization level. The work was performed as part of the mouse ENCODE project. As part of the mouse ENCODE project Mike Snyder and colleagues studied the genome-wide transcription factor (TF) occupancy repertoires, associated epigenetic signals, and TF co-association patterns in mice and humans to broaden our understanding of the evolution of gene regulation mechanisms in mammals. The results indicate that although many aspects of TF occupied sequences are conserved in both species, the extent to which orthologous DNA segments are bound by orthologous TFs in human and mouse varies both among TFs and with genomic location. Importantly, TF occupied sequences with conserved occupancy tend to be pleiotropic; they are also enriched for single nucleotide variants (SNVs) that are known to have regulatory potential or are associated with known phenotypes. As part of the mouse ENCODE project, David Gilbert and colleagues study the relationship between replication timing and higher order chromatin domains in mouse and human. They find that boundaries of replication domains — domains within the genome which replicate at the same time — share a near one-to-one correlation with topology associated domains. These and other results reconcile cell-type specific sub-nuclear compartmentalization with developmentally stable chromosome domains and offer a unified model for large-scale chromosome structure and function.

Quality versus accuracy: result of a reanalysis of protein-binding microarrays from the DREAM5 challenge by using BayesPI2 including dinucleotide interdependence.

BackgroundComputational modeling transcription factor (TF) sequence specificity is an important research topic in regulatory genomics. A systematic comparison of 26 algorithms to learn TF-DNA binding specificity in in vitro protein-binding microarray (PBM) data was published recently, but the quality of those examined PBMs was not evaluated completely.ResultsHere, new quality-control parameters such as principal component analysis (PCA) ellipse is proposed to assess the data quality for either single or paired PBMs. Additionally, a biophysical model of TF-DNA interactions including adjacent dinucleotide interdependence was implemented in a new program - BayesPI2, where sparse Bayesian learning and relevance vector machine are used to predict unknown model parameters. Then, 66 mouse TFs from the DREAM5 challenge were classified into two groups (i.e. good vs. bad) based on the paired PBM quality-control parameters. Subsequently, computational methods to model TF sequence specificity were evaluated between the two groups.ConclusionResults indicate that both the algorithm performance and the predicted TF-binding energy-level of a motif are significantly influenced by PBM data quality, where poor PBM data quality is linked to specific protein domains (e.g. C2H2 DNA-binding domain). Especially, the new dinucleotide energy-dependent model (BayesPI2) offers great improvement in testing prediction accuracy over the simple energy-independent model, for at least 21% of analyzed the TFs.Electronic supplementary materialThe online version of this article (doi:10.1186/1471-2105-15-289) contains supplementary material, which is available to authorized users.

A comparative analysis of transcription factor binding models learned from PBM, HT-SELEX and ChIP data

Understanding gene regulation is a key challenge in today's biology. The new technologies of protein-binding microarrays (PBMs) and high-throughput SELEX (HT-SELEX) allow measurement of the binding intensities of one transcription factor (TF) to numerous synthetic double-stranded DNA sequences in a single experiment. Recently, Jolma et al. reported the results of 547 HT-SELEX experiments covering human and mouse TFs. Because 162 of these TFs were also covered by PBM technology, for the first time, a large-scale comparison between implementations of these two in vitro technologies is possible. Here we assessed the similarities and differences between binding models, represented as position weight matrices, inferred from PBM and HT-SELEX, and also measured how well these models predict in vivo binding. Our results show that HT-SELEX- and PBM-derived models agree for most TFs. For some TFs, the HT-SELEX-derived models are longer versions of the PBM-derived models, whereas for other TFs, the HT-SELEX models match the secondary PBM-derived models. Remarkably, PBM-based 8-mer ranking is more accurate than that of HT-SELEX, but models derived from HT-SELEX predict in vivo binding better. In addition, we reveal several biases in HT-SELEX data including nucleotide frequency bias, enrichment of C-rich k-mers and oligos and underrepresentation of palindromes.

Predicting tissue specific transcription factor binding sites.

BackgroundStudies of gene regulation often utilize genome-wide predictions of transcription factor (TF) binding sites. Most existing prediction methods are based on sequence information alone, ignoring biological contexts such as developmental stages and tissue types. Experimental methods to study in vivo binding, including ChIP-chip and ChIP-seq, can only study one transcription factor in a single cell type and under a specific condition in each experiment, and therefore cannot scale to determine the full set of regulatory interactions in mammalian transcriptional regulatory networks.ResultsWe developed a new computational approach, PIPES, for predicting tissue-specific TF binding. PIPES integrates in vitro protein binding microarrays (PBMs), sequence conservation and tissue-specific epigenetic (DNase I hypersensitivity) information. We demonstrate that PIPES improves over existing methods on distinguishing between in vivo bound and unbound sequences using ChIP-seq data for 11 mouse TFs. In addition, our predictions are in good agreement with current knowledge of tissue-specific TF regulation.ConclusionsWe provide a systematic map of computationally predicted tissue-specific binding targets for 284 mouse TFs across 55 tissue/cell types. Such comprehensive resource is useful for researchers studying gene regulation.

Programming of human umbilical cord mesenchymal stem cells in vitro to promote pancreatic gene expression

Human umbilical cord mesenchymal stem cells (HUMSCs) are candidates for tissue engineering and may potentially be used for transdifferentiation into pancreatic endocrine cells. The adenoviral vector is effective in transducing genes into stem cells that are refractory to gene delivery by non‑viral approaches. qPCR was used to detect the pancreatic endogenous gene expression of HUMSCs transfected by islet cell-specific transcription factors (TFs). In the present study, using adenoviruses, the mouse TFs, pancreatic and duodenal homeobox 1 (pdx1), V-maf musculoaponeurotic fibrosarcoma oncogene homolog A (mafa) and class B basic helix‑loop‑helix factor neurogenin 3 (ngn3), which are essential for pancreatic cell development, were introduced into HUMSCs to assess the expression of the pancreatic genes, glucagon, pdx1 and nk2 homeobox 2 (nkx2.2). When pdx1, mafa and ngn3 were cotransduced into HUMSCs, the expression of glucagon increased by 21‑fold at days 3 and 7 following transduction, while the endogenous pdx1 gene expression was increased by 15‑fold at day 3 and decreased by 70% at day 7. When mafa and ngn3 were cotransduced into HUMSCs, there was a 5‑fold increase in pdx1 gene expression at day 7, but no activation was observed at day 3. When mafa alone was introduced into HUMSCs, the pdx1 gene expression was elevated by 6‑fold at day 3 and decreased by 3‑fold at day 7. Transduction of ngn3 alone into HUMSCs induced nkx2.2 gene expression at day 3 but the expression levels were decreased at day 7. However, when pdx1 and ngn3 were cotransduced into HUMSCs, the expression levels of glucagon, pdx1 and nks2.2 were all lower than those observed with pdx1 or ngn3 transduction alone. These results suggested that the transduction of pdx1, mafa and ngn3 genes into HUMSCs induced the expression of the pancreatic genes, glucagon, pdx‑1 and nkx2.2, and that the expression was time dependent. In addition, different combinations of the TFs may demonstrate synergistic or antagonistic effects. This data may be beneficial for guiding future studies obtaining mature pancreatic endocrine cells from HUMSCs.

Massively parallel measurements of molecular interaction kinetics on a microfluidic platform

Quantitative biology requires quantitative data. No high-throughput technologies exist capable of obtaining several hundred independent kinetic binding measurements in a single experiment. We present an integrated microfluidic device (k-MITOMI) for the simultaneous kinetic characterization of 768 biomolecular interactions. We applied k-MITOMI to the kinetic analysis of transcription factor (TF)-DNA interactions, measuring the detailed kinetic landscapes of the mouse TF Zif268, and the yeast TFs Tye7p, Yox1p, and Tbf1p. We demonstrated the integrated nature of k-MITOMI by expressing, purifying, and characterizing 27 additional yeast transcription factors in parallel on a single device. Overall, we obtained 2,388 association and dissociation curves of 223 unique molecular interactions with equilibrium dissociation constants ranging from 2 × 10(-6) M to 2 × 10(-9) M, and dissociation rate constants of approximately 6 s(-1) to 8.5 × 10(-3) s(-1). Association rate constants were uniform across 3 TF families, ranging from 3.7 × 10(6) M(-1) s(-1) to 9.6 × 10(7) M(-1) s(-1), and are well below the diffusion limit. We expect that k-MITOMI will contribute to our quantitative understanding of biological systems and accelerate the development and characterization of engineered systems.

The Transcription Factor Encyclopedia

Here we present the Transcription Factor Encyclopedia (TFe), a new web-based compendium of mini review articles on transcription factors (TFs) that is founded on the principles of open access and collaboration. Our consortium of over 100 researchers has collectively contributed over 130 mini review articles on pertinent human, mouse and rat TFs. Notable features of the TFe website include a high-quality PDF generator and web API for programmatic data retrieval. TFe aims to rapidly educate scientists about the TFs they encounter through the delivery of succinct summaries written and vetted by experts in the field. TFe is available at http://www.cisreg.ca/tfe.

Theoretical and empirical quality assessment of transcription factor-binding motifs

Position-specific scoring matrices (PSSMs) are routinely used to predict transcription factor (TF)-binding sites in genome sequences. However, their reliability to predict novel binding sites can be far from optimum, due to the use of a small number of training sites or the inappropriate choice of parameters when building the matrix or when scanning sequences with it. Measures of matrix quality such as E-value and information content rely on theoretical models, and may fail in the context of full genome sequences. We propose a method, implemented in the program ‘matrix-quality’, that combines theoretical and empirical score distributions to assess reliability of PSSMs for predicting TF-binding sites. We applied ‘matrix-quality’ to estimate the predictive capacity of matrices for bacterial, yeast and mouse TFs. The evaluation of matrices from RegulonDB revealed some poorly predictive motifs, and allowed us to quantify the improvements obtained by applying multi-genome motif discovery. Interestingly, the method reveals differences between global and specific regulators. It also highlights the enrichment of binding sites in sequence sets obtained from high-throughput ChIP-chip (bacterial and yeast TFs), and ChIP–seq and experiments (mouse TFs). The method presented here has many applications, including: selecting reliable motifs before scanning sequences; improving motif collections in TFs databases; evaluating motifs discovered using high-throughput data sets.

High resolution models of transcription factor-DNA affinities improve in vitro and in vivo binding predictions.

Accurately modeling the DNA sequence preferences of transcription factors (TFs), and using these models to predict in vivo genomic binding sites for TFs, are key pieces in deciphering the regulatory code. These efforts have been frustrated by the limited availability and accuracy of TF binding site motifs, usually represented as position-specific scoring matrices (PSSMs), which may match large numbers of sites and produce an unreliable list of target genes. Recently, protein binding microarray (PBM) experiments have emerged as a new source of high resolution data on in vitro TF binding specificities. PBM data has been analyzed either by estimating PSSMs or via rank statistics on probe intensities, so that individual sequence patterns are assigned enrichment scores (E-scores). This representation is informative but unwieldy because every TF is assigned a list of thousands of scored sequence patterns. Meanwhile, high-resolution in vivo TF occupancy data from ChIP-seq experiments is also increasingly available. We have developed a flexible discriminative framework for learning TF binding preferences from high resolution in vitro and in vivo data. We first trained support vector regression (SVR) models on PBM data to learn the mapping from probe sequences to binding intensities. We used a novel -mer based string kernel called the di-mismatch kernel to represent probe sequence similarities. The SVR models are more compact than E-scores, more expressive than PSSMs, and can be readily used to scan genomics regions to predict in vivo occupancy. Using a large data set of yeast and mouse TFs, we found that our SVR models can better predict probe intensity than the E-score method or PBM-derived PSSMs. Moreover, by using SVRs to score yeast, mouse, and human genomic regions, we were better able to predict genomic occupancy as measured by ChIP-chip and ChIP-seq experiments. Finally, we found that by training kernel-based models directly on ChIP-seq data, we greatly improved in vivo occupancy prediction, and by comparing a TF's in vitro and in vivo models, we could identify cofactors and disambiguate direct and indirect binding.