If you can not find your poster below that probably means you have not yet confirmed you will be attending ISMB 2014. To confirm your poster find the poster acceptence email there will be a confirmation link. Click on it and follow the instructions.

Category 'E'- Epigenetics'
Poster - E01
Complex Network Mechanisms of Transcriptional Activation

Istvan Ladunga, University of Nebraska, United States
Jiang Shu, University of Nebraska, United States
Donald Weeks, University of Nebraska,
Short Abstract: Higher eukaryotes regulate transcription by large complexes that remain poorly characterized due to false positive/negative ChIP-Seq observations. We overcome these issues by analyzing several thousand highly expressed (HE) genes and characterizing chromatin signatures by regressing transcript levels from regulatory binding and by statistical comparisons. From the ENCODE experiments, we identified novel specific variants of the pre-initiation complex and the transcriptional machinery.  Based on binding sites of individual proteins, generalized linear mixed models, support vector machines, and random forests cannot predict mRNA levels very accurately. Predictions were improved by jointly binding heterodimers and -trimers of regulators. HE genes are enriched in ~100 dimers and ~220 trimers including PolII complexes with TAF1, SP1, YY1, etc.  Preferential binding of “general agents” such as TAF1 to HE genes indicates unexpected specificity. Also, several factors like SP1, were implicated in DNA repair, histone modifications, immune response, differentiation, etc., but are actually widespread in HE genes. The enrichment of HE genes in multi-loop domains of embryonic stem cells may indicate novel transcriptional factories, highly efficient structures that intensify transcription. In conclusion, we identified several joint binding events of transcription factors with novel, quantitative functions. Our results contribute to the re-annotation of transcription factor function in lymphoblastoid and human embryonic stem cells.
Poster - E02
Single Molecule-level Characterization of Bacterial Methylomes

John Beaulaurier, Icahn School of Medicine at Mount Sinai, United States
Short Abstract: Genome-wide bacterial methylation analyses have not been possible until the very recent advent of Single Molecule Real-Time sequencing (SMRTseq) technique that can detect N6-methyladenine (6mA) and 4-methylcytosines (4mC), the two major types of methylations in the bacterial world in addition to 5-methylcytosines (5mC). The methylomes of a number of pathogenic and commensal bacteria are being characterized, which have revealed unprecedented diversity of bacterial methylome. Existing characterizations mainly done at population-level cannot effectively resolve epigenetic heterogeneity that often exists both within each bacterial cell and among different cells in a single bacterial colony, which empowers bacteria to better adapt to different environmental and host conditions. We present the first method for comprehensive single molecule-level characterization of bacterial methylomes, and demonstrate the substantially higher resolution it provides for nine bacterial strains, revealing different types of heterogeneity and biological insights. The proposed method can enhance the analyses and interpretation of a fast growing number of bacterial methylomes, towards deeper insights into the diverse roles of methylations in bacterial virulence, host adaption and antibiotic resistance.
Poster - E03
Spectacle: Faster and more accurate chromatin state annotation using spectral learning

Kevin Chen, Rutgers, United States
Jimin Song, Rutgers University, United States
Short Abstract: Recently, a wealth of epigenomic data has been generated by biochemical assays and next-generation sequencing (NGS) technologies. In particular, histone modification data generated by the ENCODE project and other large-scale projects show specific patterns associated with regulatory elements in the human genome.It is important to build a unified statistical model to decipher the patterns of multiple histone modifications in a cell type to annotate chromatin states such as transcription start sites, enhancers and transcribed regions rather than to map histone modifications individually to regulatory elements. Several genome-wide statistical models have been developed based on hidden Markov models (HMMs). These methods typically use the Expectation-Maximization (EM) algorithm to estimate the parameters of the model.Here we used spectral learning, a state-of-the-art parameter estimation algorithm in machine learning.We found that spectral learning plus a few (up to five) iterations of local optimization of the likelihood outperforms the standard EM algorithm.We also evaluated our software implementation called Spectacle on independent biological datasets and found that Spectacle annotated experimentally defined functional elements such as enhancers significantly better than a previous state-of-the-art method. Spectacle can be downloaded from .
Poster - E04
WARDROBE Experiment Management System : Web-based Platform for Integrated Epigenomics Analysis

Andrey Kartashov, Cincinnati Childrens Hospital Medical Center, United States
Artem Barski, Cincinnati Childrens Hospital Medical Center, United States
Short Abstract: With a growing diversity of high throughput sequencing-based epigenomics datasets and increasingly sophisticated analysis, software systems for integrated data storage, processing and visualization have become a necessity in today’s life science labs. Building upon open source software and bioinformatics tools, Wardrobe has been developed as an integrated web-based platform aiming to facilitate data analysis by biologists with no extensive experience in informatics. Wardrobe operates with MySQL, R, python, C/C++ on the back end, and with PHP, Javascript on the front end. Automated analysis pipelines include quality control, read mapping with Tophat and RPKM calculation for RNA-Seq, while mapping with Bowtie and island identification with MACS is used with ChIP-Seq and similar datasets. Graphical and table-based output is stored in the database for immediate access via web GUI. To simplify RNA-Seq and ChIP-Seq data visualization, tracks are automatically deposited to a local instance of UCSC Genome Browser. External datasets can also be automatically uploaded into Wardrobe from ftp servers, such as GEO. User managed analysis currently includes comparison of gene expression profiles between groups of samples using DESeq, comparison of ChIP-Seq enrichment using MANorm, as well as construction of ChIP-Seq tag density profiles and heatmaps for user identified gene sets. The system is modular and allows flexible integration of additional analytical tools. The current version of Wardrobe is available at
Poster - E05
MOABS: Model based analysis of bisulfite treated DNA methylation data

Deqiang Sun, Baylor Colelge of Medicine, United States
Wei Li, Baylor College of Medicine, United States
Short Abstract: Understanding the functional role of DNA methylation requires knowledge of its distribution in the genome. Bisulfite treatment followed by deep sequencing (BS-Seq) has emerged as the gold standard to study genome-wide DNA methylation at single-nucleotide resolution. While progress in next-generation sequencing allows increasingly affordable BS-seq, the resulting exponential data growth poses significant bioinformatics challenges. Here, we developed a novel bioinformatics pipeline, MOABS, to increase the speed, accuracy, statistical power and biological relevance of the BS-seq data analysis. MOABS introduces a novel strategy to combine statistical p-value and biological difference into a single metric, termed credible methylation difference (CDIF), and has enough power to detect single-CpG resolution differential methylation in small regulatory regions, such as transcription factor binding sites (TFBSs), with as low as 4-10 fold coverage. Our simulation study reveals superior performance of MOABS over other leading algorithms, such as Fisher’s exact test. Using real whole genome BS-seq data, we demonstrate that MOABS improves the detection of allele-specific DNA methylation as well as differential methylation underlying TFBSs, especially at low sequencing depth. In addition, MOABS analysis can be easily extended to more complicated scenarios, such as differential 5hmc analysis using a combination of RRBS and oxBS-seq. MOABS, as a complete, accurate and efficient solution for analysis of large scale base-resolution DNA methylation data, its source code of MOABS is available at
Poster - E06
Expression patterns and transcriptional regulation of sex-specific lincRNAs in mouse liver.

Tisha Melia, Boston University, United States
David Waxman, Boston University, United States
Short Abstract: Sex biased expression patterns characterize ~1,000 genes in mammalian liver, and impart sex differences in metabolism and disease susceptibility. The sex-dependent temporal patterns of pituitary GH secretion, pulsatile in males and continuous in females, sex-differentially activate transcription factors (TFs) and epigenetic regulators, leading to widespread sex-differences in the hepatic transcriptome. The sex-differential chromatin states are a major regulator of sex-biased liver gene expression (Mol Cell Biol 2013; PMID:23836885). We hypothesize that GH-regulated lincRNAs serve as epigenetic regulators that guide chromatin modifying enzymes to sex-specific genes. Presently, we have characterized gene structures and expression patterns for 4,961 mouse liver-expressed lincRNAs, ~60% of them novel. 247 lincRNAs showed strong sex-biased gene expression, many of which are nuclear-enriched and dependent on the temporal pattern of GH secretion for their sex-biased expression. The sex-biased lincRNAs showed strong enrichment for nearby sex-biased binding of STAT5 and FOXA2, and for male-biased binding of HNF6 and FOXA1, indicating a role for these TFs in lincRNA transcription. The sex-biased lincRNAs are also significantly enriched at sex-differential DNase I hypersensitivity sites, indicating proximal regulation of lincRNA promoters. Two distinct chromatin states, inactive and bivalent, are preferentially used to repress female-specific lincRNAs in male liver. Preliminary conservation analysis identified orthologs in rats for many liver lincRNAs, and per-base conservation analysis of lincRNA promoters showed higher conservation levels compared to protein-coding promoters and lincRNA exons, providing evidence for purifying selection. Elucidation of functional roles for sex-biased lincRNAs in gene regulation is ongoing.
Supported in part by NIH grant DK33765 (to DJW).
Poster - E07
A novel computational method for systematic analysis of chromatin profiles with applications to stem cell reprogramming in mouse

Petko Fiziev, University of California, Los Angeles, United States
Jason Ernst, University of California, Los Angeles, United States
Constantinos Chronis, University of California, Los Angeles, United States
Kathrin Plath, University of California, Los Angeles, United States
Short Abstract: Epigenetic dynamics as manifested by changes in the repertoire of post-translational histone modifications have been shown to associate with control mechanisms for the expression of genes. To systematically identify and prioritize locations in the genome with important chromatin state transitions, we developed a novel computational method for direct pairwise comparison of multiple histone marks from two cell types or conditions. The method finds chromatin differences that associate with changes in a given biological feature such as DNase I hypersensitivity or gene expression. Furthermore, it takes into account the correlation structure of multiple histone marks and, thus, is able to control for potential false positive hits. We present results from applications of our method to chromatin data sampled at several distinct stages of cellular reprogramming of mouse embryonic fibroblasts towards pluripotent stem cells. Our findings help in uncovering and understanding the major regulatory switches and events that drive this molecular process.
Poster - E08
BisuKit: A Python-based pipeline for large-scale bisulfite sequencing analysis

Jawon Song, University of Texas, Austin, United States
Matthew Vaughn, University of Texas, Austin, United States
Nathan Springer, University of Minnesota,
Short Abstract: DNA methylation is a biochemical modification of DNA where a methyl group is added onto a cytosine nucleotide. This phenomenon is known to largely affect gene expression, stem cell differentiation as well as the development of cancer cells. Therefore, understanding the relationship between the genetics and this epigenetic event is crucial. Recent advances in bisulfite sequencing technology have enabled us to analyze DNA methylation to finer resolutions with high coverage data. However, despite the fact there is a number of alignment algorithms available for bisulfite sequencing thanks to the sequencing technology, a pipeline for the downstream analysis of this large-scale bisulfite alignment data is yet to be developed. Here, we describe BisuKit, a Python-based pipeline that can be fully customized for the typical downstream analysis for bisulfite data from different species. The script can construct tiling profile of bisulfite data for visualization in genome browsers such as IGV. The pipeline also has a function to find differentially methylated regions (DMRs) directly from alignment file through wrapping existing Bioconductor packages in Python. More importantly, BisuKit offers easy implementation of new features, as existing functions were written in a modular manner. We also demonstrated its use case in large-scale data analysis using both human and maize bisulfite sequencing data.
Poster - E09
Characterizing the regulatory element landscape upon activation of B lymphocytes

Ewy Mathe, National Institute of Arthritis, Musculoskeletal, and Skin Diseases, United States
Kyong-Rim Kieffer-Kwon, National Institute of Arthritis, Musculoskeletal, and Skin Diseases, United States
Rafael Casellas, National Institute of Arthritis, Musculoskeletal, and Skin Diseases, United States
Short Abstract: Transcriptional regulation involves a complex interplay between regulatory domains (RDs), such as promoters and enhancers, that carry unique combinations of histone marks, DNA methylation, and transcription factors. Recently, by comparing the RD landscape in embryonic stem cells and activated B cells, we demonstrated that the enhancer landscape varies greatly across tissues, even for broadly transcribed genes. The underlying features of RDs upon activation of B lymphocytes is yet to be characterized.

In this study, we used DNAseI-seq to define the location of RDs in resting and activated B cells. To further complete these active regulatory landscape, we overlaid whole genome single-base resolution methylation, 5’-hydroxymethylcytosine, histone modifications, and digital footprinting onto the RDs and assessed transcription via RNA-seq. By focusing on regions that are gained and lost upon B cell activation, we were able to better understand when and how the overall increase in gene expression occurs when B cells are activated.

As expected, we find a large increase (41%) in the number of RDs, especially enhancers, upon B cell activation. This finding is consistent with the idea that RDs are gained upon activation to increase overall transcription levels. Of particular interest, we also find that while the activity in enhancers, as measured by H3K27Ac, is increased upon activation, the activity of lost enhancers is not markedly decreased. This observation suggests that the gain of enhancers is contributing to transcriptome amplification. Further characterization of these gained enhancers in terms of transcription factor recruitment and methylation will be presented here.
Poster - E10
Spectacle: Faster and more accurate chromatin state annotation using spectral learning

Jimin Song, Rutgers University, United States
Kevin Chen, Rutgers University, United States
Short Abstract: Recently, a wealth of epigenomic data has been generated by biochemical assays and next-generation sequencing (NGS) technologies. In particular, histone modification data generated by the ENCODE project and other large-scale projects show specific patterns associated with regulatory elements in the human genome.
It is important to build a unified statistical model to decipher the patterns of multiple histone modifications in a cell type to annotate chromatin states such as transcription start sites, enhancers and transcribed regions rather than to map histone modifications individually to regulatory elements.

Several genome-wide statistical models have been developed based on hidden Markov models (HMMs). These methods typically use the Expectation-Maximization (EM) algorithm to estimate the parameters of the model. Here we used spectral learning, a state-of-the-art parameter estimation algorithm in machine learning. We found that spectral learning plus a few (up to five) iterations of local optimization of the likelihood outperforms the standard EM algorithm.
We also evaluated our software implementation called Spectacle on independent biological datasets and found that Spectacle annotated experimentally defined functional elements such as enhancers significantly better than a previous state-of-the-art method.

Spectacle can be downloaded from
Poster - E11
Genome wide epigenetic profiling of human spermatozoa demonstrate altered regulating elements in obese population

Kui Qian, University of Copenhagen, Denmark
Soetkin Versteyhe, University of Copenhagen, Denmark
Ida Donkin, University of Copenhagen, Denmark
Lars Ingerslev, University of Copenhagen, Denmark
Francisc Kantor, University of Copenhagen, Denmark
Niels Jørgensen, Copenhagen University Hospital, Denmark
Juleen Zierath, University of Copenhagen, Denmark
Romain Barres, University of Copenhagen, Denmark
Short Abstract: In human, epidemiological evidence indicates a role for paternal inheritance in the predisposition of Obesity and Type 2 Diabetes (T2D), which imply that gametes carried by male gametes. To find specific epigenetic contents that may predispose to T2D, we studied spermatozoa from obese human by next generation sequencing (NGS).
We comprehensively profiled the epigenetic content of ultra pure, motile spermatozoa fractions, collected from 11 Obese (BMI>30) and 16 lean (BMI 20-25) men in reproductive age (20-40 y). Several NGS technical platforms were applied, including small RNA sequencing, MNase-Seq for histone retention and, MDB-Seq for DNA methylation.
Bioinformatics analysis revealed a specific epigenetic pattern in spermatozoa form obese subjects. While MNase-seq revealed no differential histone positioning, a subset of small RNAs were differentially expressed, including known and novel small RNA species. Integrated analysis of putative targets revealed possible role in regulating embryologic development, cell differentiation and metabolism.
Our study shows that obese human carry a distinct epigenetic profile. These epigenetic information have the potential to be early modulators of gene expression in the fertilized oocyte and thus alter embryological development and participate in the predisposition of obesity and T2D.
Poster - E12
Evaluation of Methods to Test Genomic Regions for Differential DNA Methylation in Bisulfite Sequencing Data

Hans-Ulrich Klein, University of Münster, Germany
Katja Hebestreit, Stanford University, United States
Short Abstract: DNA methylation is an epigenetic mechanism that is known to regulate stem cell differentiation and to play a role in diseases. Bisulfite sequencing has been frequently used to quantify CpG methylation. Given the fact that only a relatively small fraction of all genomic CpGs shows differential methylation levels across a large number of cell types, targeted approaches are cost efficient alternatives for studying the methylome.
Testing target regions instead of single CpGs decreases the number of tests remarkably and so potentially increases the power. We evaluated, adapted and compared seven methods for testing the null hypothesis that at least one CpG is differentially methylated in a given genomic region that may contain a large number of CpGs. The quantitative comparison was conducted on a targeted bisulfite data set of biological replicates after incorporating artificial differentially methylated regions of different lengths and methylation differences. Receiver operator curves (ROC) were then calculated to assess the different methods.
Global test, BiSeq and limma were applied after smoothing the methylation levels and outperformed three different log-linear models that model raw read counts. Biseq was superior to limma demonstrating that the beta distribution is an appropriate choice for modeling methylation levels. COHCAP, an analysis pipeline designed for methylation data, performed similar to limma. In summary, BiSeq and global test are preferable compared to frequently used log-linear models like the Poisson regression, even if the Poisson variance assumption is relaxed to account for overdispersion.
Poster - E13
Complex pattern of CAST allelic expression in bovine muscle tissue

Adhemar Zerlotini, EMBRAPA, Brazil
M. M. Souza, Universidade Federal de São Carlos, Brazil
A. Zerlotini, Embrapa Informática Agropecuária, Brazil
P. C. Tizioto, Universidade Federal de São Carlos, Brazil
P. S. N. Oliveira, Universidade Federal de São Carlos, Brazil
A. L. Somavilla, Universidade Estadual Paulista, Brazil
F. B. Mokry, Embrapa Informática Agropecuária, Brazil
Cesar A. S. M., Universidade de São Paulo, Brazil
W. J. S. Diniz, Universidade Federal de São Carlos, Brazil
M. A. Mudadu, Embrapa Pecuária Sudeste, Brazil
S. C. M. Niciura, Embrapa Pecuária Sudeste, Brazil
L. L. Coutinho, Universidade de São Paulo, Brazil
L. C. A. Regitano, Embrapa Pecuária Sudeste, Brazil
Short Abstract: Imprinted genes have been target of many studies, mainly in human and mouse, and lately in bovines due to the importance of understanding the epigenetics mechanisms underlying important phenotypes, and the possibility of applying it in animal breeding programs in the future. Genomic DNA from 38 steers was genotyped using the Illumina BovineHD BeadChip in order to identify heterozygous individuals with known allele origin. Total mRNA from muscle was extracted and sequenced by Illumina HiScanSQ and the reads were mapped to the Bos taurus reference genome. In-house software based on the phyton module pySAM enable us assess the frequency of each nucleotide. Genome wide analysis of the SNPs is being conducted. As an example, a candidate gene to meat tenderness, CAST, was previously described showing biallelic expression by real time allelic discrimination approaches in bovine muscle. In order to confirm the CAST expression in muscle, a SNP set in exon was used to assess allelic expression by RNA-seq. A total of 9 reciprocal heterozygous had more than 30 reads for the chosen SNP. A total of 6 animals showed allele ratio C/T mean 1.02. Three animals, showed monoallelic expression, of which 2 had exclusively expression of T allele and 1 of C allele. These 3 animals had the same father then the monoallelic expression can not be associated to allele parental origin. The results suggest a complex pattern of allele expression and the SNP analyzed here probably is not the causal underlying this pattern.
Poster - E14
Modeling complex patterns of differential DNA methylation that strongly associate with gene expression changes

Christopher Schlosberg, Washington University in St. Louis, United States
Nathan VanderKraats, Washington University in St. Louis, United States
Jeffery Hiken, Washington University in St. Louis, United States
Kilian Weinberger, Washington University in St. Louis, United States
Tao Ju, Washington University in St. Louis, United States
John Edwards, Washington University in St. Louis, United States
Short Abstract: Establishment of specific patterns of DNA methylation is necessary for normal development, and aberrant methylation is frequently observed in cancer. Hypermethylation of CpG islands overlapping the transcription start site (TSS) downregulates tumor suppressor genes, thus promoting tumorigenesis. However, recent genome-wide mapping of methylation indicates only modest correlation between differential gene expression (DGE) and methylation, casting doubt on the importance of methylation in regulating DGE. In addition, complex patterns, such as CpG island-shore methylation and long hypomethylated domains, also correlate with DGE. We hypothesize that unbiased computational tools will better model complex patterns of methylation and capture strong associations between DGE and methylation. By representing methylation as continuous curves centered on a gene’s TSS and performing unsupervised clustering using Dynamic Time Warping, we enumerate complex, differential methylation signatures that highly correlate with DGE. We next trained a nearest neighbor classifier on examples of these significantly correlated signatures to identify genes that display both differential methylation and expression. Using data from the Human Epigenome Atlas, ENCODE, and eight breast cancer cell lines, our classifier significantly outperforms state-of-the-art Differentially Methylated Region (DMR)- and Support Vector Machine-based methods at identifying associated genes. By further analyzing these associated genes, we find methylation’s silencing mechanism may be signature-dependent. In breast cancer cells, we observe that methylation at the TSS does not affect transcriptional initiation, however, methylation proximal to the TSS may inhibit transcriptional elongation. The discovery of these potentially functional methylation changes will facilitate the identification of patients who may benefit from clinically-approved demethylating therapeutics.
Poster - E15
Chromatin position effects assayed by thousands of reporters integrated in parallel

Johann de Jong, The Netherlands Cancer Institute, Netherlands
Waseem Akhtar, The Netherlands Cancer Institute, Netherlands
Alexey Pindyurin, Siberian Branch of Russian Academy of Sciences, Russia
Ludo Pagie, The Netherlands Cancer Institute, Netherlands
Jeroen de Ridder, Delft University of Technology, Netherlands
Anton Berns, The Netherlands Cancer Institute, Netherlands
Lodewyk Wessels, The Netherlands Cancer Institute, Netherlands
Maarten van Lohuizen, The Netherlands Cancer Institute, Netherlands
Bas van Steensel, The Netherlands Cancer Institute, Netherlands
Short Abstract: Reporter genes integrated into the genome are a powerful tool to reveal effects of regulatory elements and local chromatin context on gene expression. However, so far such reporter assays have been of low throughput.

Here, we present results of computational analyses using a novel approach for the parallel monitoring of transcriptional activity of thousands of integrated reporter genes throughout the genome.

More than 27000 distinct reporter integrations in mouse embryonic stem cells, obtained with two different promoters, show ∼1000-fold variation in expression levels. This variation can be partially explained by association with the inner nuclear membrane, by the degree of compaction of the chromatin, and by the proximity to promoters and enhancers of endogenous genes.

Analysis also shows cooperative and inhibitory interactions between different transcription factors. Reporters landing in repeats are often active and there is a remarkable variation in the expression of reporters depending on the repeat type.
Poster - E16
Comparative epigenomics across age, sex, tissues, sample heterogeneity, and suspension state.

Angela Yen, Massachusetts Institute of Technology, United States
Short Abstract: Epigenomic datasets provide dynamic information about the activity patterns of genomic regions, which has enabled comparison of regulatory region dynamics, similar to gene expression analysis. However, while the field of systems biology has benefited from many analysis tools for distinguishing differentially-expressed genes, the field of epigenomics is currently lacking tools for studying systematic differences across epigenomic groups, which are necessary given the increasing availability of epigenomes. Here, we introduce a new method for comparing large sets of epigenomic datasets across multiple samples, which enables comparisons of groups of tissues and cell types and leverages the genome-wide nature of epigenomic datasets.

We apply this method to 127 epigenomes from the Roadmap Epigenomics project, to investigate how chromatin state varies across age, sex, source, heterogeneity, and suspension state, and to reveal the pathways associated with these changes. Remarkably, we find that different epigenomic features are maximally discriminative for different group-wise comparisons. For example, we find that between male and female samples, the primary differences involve changes in repressed and quiescent chromatin states, and are associated with gender effects and Suz12 targets, consistent with X inactivation. Further, between adult and fetal samples, differences involve promoter, transcribed, and repressed regions, at genes enriched for immune functions and associated with cancer and cell cycle pathways.

Our results suggest remarkable and consistent signatures distinguishing the epigenomic landscapes of classes of samples. Our methodology is general, should be broadly applicable for epigenomic comparisons, and provides a powerful new tool for studying epigenomic differences at the genome scale.
Poster - E17
Using Epigenomic Data to Reveal Cell-specific Regulation of Gene Expression

Duygu Ucar, Jackson Laboratory, United States
Short Abstract: The overall objective of this project is to discover cell-specific regulatory elements and their distal interactions in primary human cells and tissues by integrating reference epigenome maps (profiled by the NIH Roadmap Epigenomics Program) with other complementary datasets that contain information on chromatin interactions, gene expression, and protein binding. Our central hypothesis is that cell-specific regulators are marked by distinct genomic landscapes and these landscapes are readily amenable to computational and data-centric approaches. Our project aims to identify cell-specific promoters and enhancers in diverse human cell types through an integrative data-mining platform that enables feature extraction, classification, and feature ranking. We developed a data-mining platform that enables integrating and extracting information from diverse genome-wide datasets. With this platform we study examples of known cell-specific regulators to identify their genomic and epigenomic characteristics, for example in stem cells we studied distinct characteristics of stem cell regulators identified by RNAi screens. Our integrative method enables discovering genomic and epigenomic biomarkers that are specifically marking these cell-specific genes. This enables furthering our understanding of these cell-specific regulatory elements that are essential for cell function and disease susceptibility.
Poster - E18
HMCan – a tool to detect chromatin modifications in cancer samples using ChIP-seq data

HAITHAM ASHOOR, King Abdullah University of Science and Technology, Saudi Arabia
Vladimir Bajic, King Abdullah University of Science and Technology, Saudi Arabia
Emmanuel Barillot, Institut Curie, France
Valentina Boeva, Institut Curie, France
Short Abstract: Epigenetic changes often play an important role in cancer development. Though several tools were created to enable detection of histone marks in ChIP-seq data from normal samples, it was unclear whether these tools can be applied to ChIP-seq data generated from cancer samples. The challenge comes from the fact that cancer genomes are often characterized by frequent copy number alterations: gains and losses of large regions of chromosomal material; which may create a substantial statistical bias in the evaluation of histone mark signal enrichment resulting in underdetection or overdetection of the signal. We present HMCan (Histone Modifications in Cancer), a recently published tool (Ashoor et al.), specially developed to analyze histone modification ChIP-seq data produced from cancer genomes. HMCan corrects for the GC- and copy number bias and then applies Hidden Markov Models (HMMs) to detect the signal in the normalized profile. On simulated data, HMCan outperformed several commonly used tools such as CCAT, MACS and SICER. Of note, MACS and SICER were biased towards regions of copy number gain, while CCAT and HMCan did not demonstrate such a bias. HMCan also showed superior results on a ChIP-seq dataset generated for the repressive histone mark H3K27me3 in a bladder cancer cell line, where predictions matched well with experimental data (qPCR validated regions).
Ashoor,H. et al. (2013) HMCan: a method for detecting chromatin modifications in cancer samples using ChIP-seq data. Bioinformatics, 29, 2979–2986.
Poster - E19
De-novo inference of long-range enhancer-promoter interactions

Jianrong Wang, Massachusetts Institute of Technology, United States
Anshul Kundaje, Massachusetts Institute of Technology, United States
Manolis Kellis, Massachusetts Institute of Technology, United States
Short Abstract: Long-range interactions between distal enhancers and their target genes play a pivotal role in cell-type specific regulation of gene expression. Chromatin state maps in diverse cell-types have provided us a comprehensive annotation of distal enhancers. However, experimental datasets to characterize in vivo long-range regulatory interactions are limited to a small number of cell types and computational linking methods often rely on assumptions such as genomic proximity, resulting in the lack of a comprehensive and accurate map of context-specific target genes of distal enhancers. We developed a novel probabilistic model to computationally infer putative target genes of cell-type specific enhancers based on the associated chromatin state and gene expression dynamics across 56 diverse human cell-types. We automatically discover co-activated transcriptional and enhancer modules that were strongly enriched for lineage specific functional annotations and biochemical pathways; as well as the complex, non-linear, dynamic cell-type specific interactions between transcriptional and enhancer modules. The resulting model showed significant improvements in prediction of transcriptional responses. The accuracy and cell-type specificity of our predicted links were further validated by experimental chromatin interaction data (including Hi-C and ChIA-PET datasets) and eQTL annotations. Analysis of the genomic sequence underlying linked enhancers and promoters revealed significant co-enrichment of specific combinations of transcription factor binding motifs, providing specific hypotheses about long-range regulatory interactions between transcription factors. Thus, our model provides novel insights into the complex interactions between enhancers and their target genes and the dynamic re-organization of long-range regulatory interactions that mediate cell-type specific transcriptional responses.
Poster - E20
Genome-wide differential variability methylation analysis on CRC tumors and cell lines identifies candidate DNA methylation markers of response to MEK Inhibitors

Sara Dempster, AstraZeneca, United States
Andreas Schlicker, The Netherlands Cancer Institute, Netherlands
Lodewyk Wessels, The Netherlands Cancer Institute, Netherlands
Neil Gibson, AstraZeneca, United Kingdom
Garry Beran, AstraZeneca, United Kingdom
Christine Chresta, AstraZeneca, United Kingdom
Mandy Ng, AstraZeneca, United Kingdom
Short Abstract: DNA methylation changes have recently been established as a driving force of cancer development, but the utility of DNA methylation for predicting response to targeted cancer therapies hasn’t been established. To address the question in colorectal cancer (CRC), using the Infinium450 array, we generated a novel genome-wide DNA methylation dataset reporting methylation status at more than 450,000 methylation sites in 49 CRC cell lines, 33 CRC tumors and 10 normal colon samples. We found that DNA methylation patterns at a large number of methylation probes are highly conserved in normal samples while tumors and cancer cell lines exhibit variable methylation patterns. We identify 2469 genomic loci referred to as SOMS which exhibited significantly higher DNA methylation variance in tumors and cell lines compared to normal colon tissue. We integrated pharmacological response data with DNA methylation data for the cell lines, and observed statistically significant associations between methylation status at five loci and response to targeted MEK inhibitors. We refer to these loci as cancer response associated methylation sites (CRAMSs). Expression levels of the genes comprising these loci were correlated with the respective methylation levels, providing evidence for a functional impact of the methylation differences between different drug response subtypes. Furthermore, expression levels of these same CRAMS containing genes were also predictive of response to MEK inhibitors in the Cancer Cell Line Encyclopedia pan tumor-type panel providing an independent validation. Our results suggest that changes in DNA methylation can serve as biomarker for response.
Poster - E21
Characterization of enhancer gene interactions using DNaseI and gene expression data cross 110 cell types

Pouya Kheradpour, Massachusetts Institute of Technology, United States
Roadmap Epigenomics Consortium, , United States
Manolis Kellis, Massachusetts Institute of Technology, United States
Short Abstract: Recent efforts to characterize diseases through genome-wide association studies and annotate the genome using ChIP-seq experiments have led to a dramatic increase in putative functional genomic regions. While most of the implicated loci have fallen outside coding regions and are thought to be regulatory in nature, efforts to link these regions to their target genes, thereby permitting a better understanding of their importance, has lagged considerably. Generally, experimental linking techniques only permit the interrogation of a small number of specific regions or produce a genome-wide linking at very low resolution.

We utilize DNaseI hypersensitivity and expression data from 110 cell types produced by the ENCODE and Roadmap Epigenomics projects to produce linking confidences between hypersensitive regions and nearby genes. We find that high confidence links are supported by independent datasets such as eQTL annotations. Moreover, we find that after controlling for distance several factors, including the orientation of the upstream gene and the presence of an interleaving CTCF, greatly affect the linking confidence.

In addition to predicting links between genes and nearby putative enhancers, we are interested in answering fundamental biological questions such as the number of enhancers per gene and where these enhancers fall relative to a gene's transcription start site. We address these questions through a careful analysis of the distribution of confidence scores for each gene and by performing a scaling analysis on the number of cell types.


Posters Display Schedule:

ALL Posters - set up for display on Sunday, July 13 (8:00 am - 10:00 am)

Odd Numbered poster presentations:
  • Author presentations will take place Sunday, July 13: 5:00 p.m. - 7:00 p.m.
Even Numbered poster presentations:
  • Author presentations will take place Monday, July 14: 5:45 p.m. - 7:30 p.m.
Remove All Posters - Tuesday, July 15 after 4:30 pm

Poster Sessions with Authors are held 5:45 p.m. - 7:30 p.m. on Sunday, July 13 (odd numbered posters) and 5:45 p.m. - 7:30 p.m. Monday, July 14 (even numbered posters). Authors should be prepared in advanced to be available to present on their assigned day. Please note due to the large number of posters we are unable to accept special requests for presentation dates.

Poster Display Size: When preparing accepted posters please note that your poster should not exceed the following dimensions: 46 inches wide by 45 inches high. There will be 2 posters per side on the each poster board. One poster will be an odd number and the other will be an even number. View a diagram of the the poster board in pdf format here.

Print your poster in Boston

Delegates wishing to have their poster printed in advance could use the services of Our pricing is by linear foot of poster. Pricing for ISMB attendees would be as follows:
46": $45
36": $35

View Posters By Category
Search Posters:
Poster Number Matches
Last Name
Co-Authors Contains
Abstract Contains