A supervised learning framework for chromatin loop detection in genome-wide contact maps.
Ontology highlight
ABSTRACT:
SUBMITTER: Salameh TJ
PROVIDER: S-EPMC7347923 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:MotivationThe high-throughput chromosome conformation capture (Hi-C) technique has enabled genome-wide mapping of chromatin interactions. However, high-resolution Hi-C data requires costly, deep sequencing; therefore, it has only been achieved for a limited number of cell types. Machine learning models based on neural networks have been developed as a remedy to this problem.ResultsIn this work, we propose a novel method, EnHiC, for predicting high-resolution Hi-C matrices from low-resolution input data based on a generative adversarial network (GAN) framework. Inspired by non-negative matrix factorization, our model fully exploits the unique properties of Hi-C matrices and extracts rank-1 features from multi-scale low-resolution matrices to enhance the resolution. Using three human Hi-C datasets, we demonstrated that EnHiC accurately and reliably enhanced the resolution of Hi-C matrices and outperformed other GAN-based models. Moreover, EnHiC-predicted high-resolution matrices facilitated the accurate detection of topologically associated domains and fine-scale chromatin interactions.Availability and implementationEnHiC is publicly available at https://github.com/wmalab/EnHiC.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC8382278 | biostudies-literature
Project description:The yield of cereal crops such as sorghum (Sorghum bicolor L. Moench) depends on the distribution of crop-heads in varying branching arrangements. Therefore, counting the head number per unit area is critical for plant breeders to correlate with the genotypic variation in a specific breeding field. However, measuring such phenotypic traits manually is an extremely labor-intensive process and suffers from low efficiency and human errors. Moreover, the process is almost infeasible for large-scale breeding plantations or experiments. Machine learning-based approaches like deep convolutional neural network (CNN) based object detectors are promising tools for efficient object detection and counting. However, a significant limitation of such deep learning-based approaches is that they typically require a massive amount of hand-labeled images for training, which is still a tedious process. Here, we propose an active learning inspired weakly supervised deep learning framework for sorghum head detection and counting from UAV-based images. We demonstrate that it is possible to significantly reduce human labeling effort without compromising final model performance (R 2 between human count and machine count is 0.88) by using a semitrained CNN model (i.e., trained with limited labeled data) to perform synthetic annotation. In addition, we also visualize key features that the network learns. This improves trustworthiness by enabling users to better understand and trust the decisions that the trained deep learning model makes.
| S-EPMC7706102 | biostudies-literature
Project description:We applied genome-wide profiling to successive salt-extracted fractions of micrococcal nuclease-treated Drosophila chromatin. Chromatin fractions extracted with 80 mM or 150 mM NaCl after digestion contain predominantly mononucleosomes and represent classical "active" chromatin. Profiles of these low-salt soluble fractions display phased nucleosomes over transcriptionally active genes that are locally depleted of histone H3.3 and correspond closely to profiles of histone H2Av (H2A.Z) and RNA polymerase II. This correspondence suggests that transcription can result in loss of H3.3+H2Av nucleosomes and generate low-salt soluble nucleosomes. Nearly quantitative recovery of chromatin is obtained with 600 mM NaCl; however, the remaining insoluble chromatin is enriched in actively transcribed regions. Salt-insoluble chromatin likely represents oligonucleosomes that are attached to large protein complexes. Both low-salt extracted and insoluble chromatin are rich in sequences that correspond to epigenetic regulatory elements genome-wide. The presence of active chromatin at both extremes of salt solubility suggests that these salt fractions capture bound and unbound intermediates in active processes, thus providing a simple, powerful strategy for mapping epigenome dynamics.
| S-EPMC2661814 | biostudies-literature
Project description:Current methods for whole-genome mapping of protein-DNA interactions, performed by coupling chromatin immunoprecipitation with high-throughput sequencing (ChIP-seq), require large amounts of starting materials, which precludes their application to rare cell types. Here we combine a high-sensitivity ChIP assay with a new library preparation procedure to map histone modifications in as few as 10,000 cells. We used the technique to characterize mouse hematopoietic progenitors and thereby gain insight into their developmental program.
| S-EPMC2924612 | biostudies-literature
Project description:BACKGROUND:The spatial configuration of chromosomes is essential to various cellular processes, notably gene regulation, while architecture related alterations, such as translocations and gene fusions, are often cancer drivers. Thus, eliciting chromatin conformation is important, yet challenging due to compaction, dynamics and scale. However, a variety of recent assays, in particular Hi-C, have generated new details of chromatin structure, spawning a number of novel biological findings. Many findings have resulted from analyses on the level of native contact data as generated by the assays. Alternatively, reconstruction based approaches often proceed by first converting contact frequencies into distances, then generating a three dimensional (3D) chromatin configuration that best recapitulates these distances. Subsequent analyses can enrich contact level analyses via superposition of genomic attributes on the reconstruction. But, such advantages depend on the accuracy of the reconstruction which, absent gold standards, is inherently difficult to assess. Attempts at accuracy evaluation have relied on simulation and/or FISH imaging that typically features a handful of low resolution probes. While newly advanced multiplexed FISH imaging offers possibilities for refined 3D reconstruction accuracy evaluation, availability of such data is limited due to assay complexity and the resolution thereof is appreciably lower than the reconstructions being assessed. Accordingly, there is demand for new methods of reconstruction accuracy appraisal. RESULTS:Here we explore the potential of recently proposed stationary distributions, hereafter StatDns, derived from Hi-C contact matrices, to serve as a basis for reconstruction accuracy assessment. Current usage of such StatDns has focussed on the identification of highly interactive regions (HIRs): computationally defined regions of the genome purportedly involved in numerous long-range intra-chromosomal contacts. Consistent identification of HIRs would be informative with respect to inferred 3D architecture since the corresponding regions of the reconstruction would have an elevated number of k nearest neighbors (kNNs). More generally, we anticipate a monotone decreasing relationship between StatDn values and kNN distances. After initially evaluating the reproducibility of StatDns across replicate Hi-C data sets, we use this implied StatDn - kNN relationship to gauge the utility of StatDns for reconstruction validation, making recourse to both real and simulated examples. CONCLUSIONS:Our analyses demonstrate that, as constructed, StatDns do not provide a suitable measure for assessing the accuracy of 3D genome reconstructions. Whether this is attributable to specific choices surrounding normalization in defining StatDns or to the logic underlying their very formulation remains to be determined.
| S-EPMC7041182 | biostudies-literature
Project description:The three-dimensional configuration of DNA is integral to all nuclear processes in eukaryotes, yet our knowledge of the chromosome architecture is still limited. Genome-wide chromosome conformation capture studies have uncovered features of chromatin organization in cultured cells, but genome architecture in human tissues has yet to be explored. Here, we report the most comprehensive survey to date of chromatin organization in human tissues. Through integrative analysis of chromatin contact maps in 21 primary human tissues and cell types, we find topologically associating domains highly conserved in different tissues. We also discover genomic regions that exhibit unusually high levels of local chromatin interactions. These frequently interacting regions (FIREs) are enriched for super-enhancers and are near tissue-specifically expressed genes. They display strong tissue-specificity in local chromatin interactions. Additionally, FIRE formation is partially dependent on CTCF and the Cohesin complex. We further show that FIREs can help annotate the function of non-coding sequence variants.
| S-EPMC5478386 | biostudies-literature
Project description:We report the application of single-molecule-based sequencing technology for high-throughput profiling of histone modifications in mammalian cells. By obtaining over four billion bases of sequence from chromatin immunoprecipitated DNA, we generated genome-wide chromatin-state maps of mouse embryonic stem cells, neural progenitor cells and embryonic fibroblasts. We find that lysine 4 and lysine 27 trimethylation effectively discriminates genes that are expressed, poised for expression, or stably repressed, and therefore reflect cell state and lineage potential. Lysine 36 trimethylation marks primary coding and non-coding transcripts, facilitating gene annotation. Trimethylation of lysine 9 and lysine 20 is detected at satellite, telomeric and active long-terminal repeats, and can spread into proximal unique sequences. Lysine 4 and lysine 9 trimethylation marks imprinting control regions. Finally, we show that chromatin state can be read in an allele-specific manner by using single nucleotide polymorphisms. This study provides a framework for the application of comprehensive chromatin profiling towards characterization of diverse mammalian cell populations.
| S-EPMC2921165 | biostudies-literature
Project description:BACKGROUND:In the human genome, 98% of DNA sequences are non-protein-coding regions that were previously disregarded as junk DNA. In fact, non-coding regions host a variety of cis-regulatory regions which precisely control the expression of genes. Thus, Identifying active cis-regulatory regions in the human genome is critical for understanding gene regulation and assessing the impact of genetic variation on phenotype. The developments of high-throughput sequencing and machine learning technologies make it possible to predict cis-regulatory regions genome wide. RESULTS:Based on rich data resources such as the Encyclopedia of DNA Elements (ENCODE) and the Functional Annotation of the Mammalian Genome (FANTOM) projects, we introduce DECRES based on supervised deep learning approaches for the identification of enhancer and promoter regions in the human genome. Due to their ability to discover patterns in large and complex data, the introduction of deep learning methods enables a significant advance in our knowledge of the genomic locations of cis-regulatory regions. Using models for well-characterized cell lines, we identify key experimental features that contribute to the predictive performance. Applying DECRES, we delineate locations of 300,000 candidate enhancers genome wide (6.8% of the genome, of which 40,000 are supported by bidirectional transcription data), and 26,000 candidate promoters (0.6% of the genome). CONCLUSION:The predicted annotations of cis-regulatory regions will provide broad utility for genome interpretation from functional genomics to clinical applications. The DECRES model demonstrates potentials of deep learning technologies when combined with high-throughput sequencing data, and inspires the development of other advanced neural network models for further improvement of genome annotations.
| S-EPMC5984344 | biostudies-literature
Project description:RNA is a critical component of chromatin in eukaryotes, both as a product of transcription, and as an essential constituent of ribonucleoprotein complexes that regulate both local and global chromatin states. Here, we present a proximity ligation and sequencing method called Chromatin-Associated RNA sequencing (ChAR-seq) that maps all RNA-to-DNA contacts across the genome. Using Drosophila cells, we show that ChAR-seq provides unbiased, de novo identification of targets of chromatin-bound RNAs including nascent transcripts, chromosome-specific dosage compensation ncRNAs, and genome-wide trans-associated RNAs involved in co-transcriptional RNA processing.
| S-EPMC5962340 | biostudies-literature
Project description:Genome-wide association studies (GWAS) associate single nucleotide polymorphisms (SNPs) to complex phenotypes. Most human SNPs fall in non-coding regions and are likely regulatory SNPs, but linkage disequilibrium (LD) blocks make it difficult to distinguish functional SNPs. Therefore, putative functional SNPs are usually annotated with molecular markers of gene regulatory regions and prioritized with dedicated prediction tools. We integrated associated SNPs, LD blocks and regulatory features into a supervised model called TAGOOS (TAG SNP bOOSting) and computed scores genome-wide. The TAGOOS scores enriched and prioritized unseen associated SNPs with an odds ratio of 4.3 and 3.5 and an area under the curve (AUC) of 0.65 and 0.6 for intronic and intergenic regions, respectively. The TAGOOS score was correlated with the maximal significance of associated SNPs and expression quantitative trait loci (eQTLs) and with the number of biological samples annotated for key regulatory features. Analysis of loci and regions associated to cleft lip and human adult height phenotypes recovered known functional loci and predicted new functional loci enriched in transcriptions factors related to the phenotypes. In conclusion, we trained a supervised model based on associated SNPs to prioritize putative functional regions. The TAGOOS scores, annotations and UCSC genome tracks are available here: https://tagoos.readthedocs.io.
| S-EPMC6698643 | biostudies-literature