Predicting Genome Architecture: Challenges and Solutions

Belokopytova, Polina; Fishman, Veniamin

doi:10.3389/fgene.2020.617202

“…To explain these differences, one should consider the diversity of DNA sequences. There are several computational methods allowing the prediction of specific epigenetic properties (Chen et al 2021;Avsec et al 2021), chromatin packaging (P. Belokopytova and Fishman 2021), or expression levels (Avsec et al 2021) based on the DNA sequences. Among these, the prediction of gene expression changes caused by sequence variations is especially interesting, because such predictions enable clinical interpretation of non-coding variants in the human genome (Avsec et al 2021).…”

mentioning

confidence: 99%

Cell type-specific interpretation of noncoding variants using deep learning-based methods

Sindeeva¹,

Chekanov²,

Avetisian³

et al. 2022

Preprint

Self Cite

1

2

0

View full text Add to dashboard Cite

Interpretation of non-coding genomic variants is one of the most important challenges in human genetics. Machine learning methods have emerged recently as a powerful tool to solve this problem. State-of-the-art approaches allow prediction of transcriptional and epigenetic effects caused by non-coding mutations. However, these approaches require specific experimental data for training and can not generalize across cell types where required features were not experimentally measured. We show here that available epigenetic characteristics of human cell types are extremely sparse, limiting those approaches that rely on specific epigenetic input. We propose a new neural network architecture, DeepCT, which can learn complex interconnections of epigenetic features and infer unmeasured data from any available input. Furthermore, we show that DeepCT can learn cell type-specific properties, build biologically meaningful vector representations of cell types and utilize these representations to generate cell type-specific predictions of the effects of non-coding variations in the human genome.

show abstract

“…Identification of chromatin states, local density of epigenetic marks, long-range chromatin contacts and histone modification patterns has proven relevant for studying and interpreting regulatory regions, cell specific activity and disease-associated patterns. To this end, many ML and DL techniques have been applied to define cell type-specific profiles of DNA methylation (or methylomes) and histone modifications, classify chromatin regions into active and repressed states and, more recently, classify tumour types based on high-throughput methylome data and predict 3D genome folding [206] , [207] , [208] .…”

Section: Ai Applications In Functional Genomicsmentioning

confidence: 99%

AI applications in functional genomics

Caudai

¹

,

Galizia

²

,

Geraci

³

et al. 2021

Computational and Structural Biotechnology Journal

View full text Add to dashboard Cite

show abstract

“…1B). Protein ensembles regulating higher-order chromatin structure include polycomb repressor complexes PRC1 and -2, and the transcriptional repressor CTCF, also known as 11-zinc finger protein or CCCTC-binding factor transcription factor, which together with cohesins shape chromatin architecture [55]. For example, CTCF bin-▶Fig.…”

Section: Three-dimensional Chromatin Organizationmentioning

confidence: 99%

Epigenetics and Noncoding RNA – Principles and Clinical Impact

Kornak

¹

,

Bischof

²

,

Ebert

³

et al. 2021

Osteologie

1

0

View full text Add to dashboard Cite

Genetics studies the inheritance of genetic information encoded by the base pair sequence and its variants. Sequence variants can have severe consequences as seen in genetically inherited diseases (e. g. osteogenesis Imperfecta, hypophosphatasia). On the other hand, epigenetics deals with inherited and dynamically reversible modifications of chromatin without changing the base pair sequence, resulting in a change in phenotype without a change in genotype. These modifications primarily exert their effects by influencing gene expression. Initially, the definition of epigenetics exclusively comprised inherited changes that persist across several generations without changes in the DNA sequence. This definition has been extended to include also dynamic and partially reversible changes that occur more short-term. These gene modulatory effects introduce new levels of complexity and are crucial determinants of cell fate and organismal development. With its length of approximately two meters, human DNA has to be compacted to fit into the nuclei and fulfill its functions. DNA is wrapped around histone octamers into so-called nucleosomes. DNA, histones, and other DNA-associated proteins together form what is called chromatin. DNA packaging is achieved by variable degrees of chromatin condensation depending on cell type and context. Epigenetic transcriptional regulation modifies the affinity and accessibility of cis-regulatory elements (CREs) for transcription factors and the basic transcriptional machinery and governs interaction between CREs. CREs include promoters, enhancers, silencers, and insulators and are potent modulators of gene expression impacting core cell biological processes such as proliferation and differentiation. Chromatin looping and remodeling by differential covalent modifications of DNA (e. g., methylation or hydroxylation) and histone tails (e. g., acetylation or methylation) elicit fundamental changes in CRE accessibility, thus impacting gene expression. Chromatin looping depends on a specialized machinery including cohesins. Chromatin modifications are mediated by specific enzymes like DNA methylases (DNMTs), histone-modifying enzymes, like histone methyl- and acetyltransferases (KMTs, HATs/KATs), and histone demethylases and deacetylases (KDMs, HDACs). It becomes increasingly evident that epigenetic (dys)regulation plays a decisive role in physiology and pathophysiology, impacting many age-related diseases like cancer and degenerative pathologies (e. g., osteoporosis, Alzheimer’s, or Parkinson’s) in a significant fashion. Recently, small-molecule inhibitors of chromatin-modifying enzymes (e. g., vorinostat) have been identified and successfully introduced in therapy. Significant progress in high-throughput sequencing technologies and big data analysis has broadened our understanding of noncoding (nc) RNAs and DNA sequence regions in (post-)transcriptional regulation and disease development. Among ncRNAs that play vital roles in gene expression are micro- (miRs) and long noncoding RNAs (lncRNAs; e. g., XIST or HOTAIR). By interacting with the coding genome, these RNAs modulate important genetic programs. Interfering RNAs can, for example, enhance the post-transcriptional degradation of transcripts, altering their translation, or assist in the recruitment of chromatin-modifying enzymes to regulate transcription. They can also be packaged into extracellular vesicles as cargo and thus deliver critical information to the microenvironment or even systemically to distant tissues. Therefore, ncRNAs represent a novel playground for therapeutical investigations and supplement epigenetic mechanisms of gene regulation while being subject to epigenetic regulation themselves. Last but not least, dysregulated ncRNAs can also propagate disease. Until recently, the detection of epigenetic phenomena necessitated invasive diagnostic interventions. However, with the arrival of so-called “liquid biopsies” an analysis of circulating cell-free DNA fragments (cfDNA) and RNAs as well as vesicle-packed RNAs through minimal invasively drawn blood samples can be obtained. Such “fragmentomics” and RNAomics approaches on peripheral blood will ultimately serve as diagnostic tools for personalized clinical interventions.

show abstract

Predicting Genome Architecture: Challenges and Solutions

Cited by 33 publications

References 134 publications

Cell type-specific interpretation of noncoding variants using deep learning-based methods

Cell type-specific interpretation of noncoding variants using deep learning-based methods

AI applications in functional genomics

Epigenetics and Noncoding RNA – Principles and Clinical Impact

Contact Info

Product

Resources

About