Unraveling the three-dimensional genome structure using machine learning
3
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
The study of chromatin interactions has advanced considerablywith technologies such as high-throughput chromosome conformationcapture (Hi-C) sequencing, providing a genome-wide viewof physical interactions within the nucleus. These techniqueshave revealed the existence of hierarchical chromatin structuressuch as compartments, topologically associating domains (TADs),and chromatin loops, which are crucial in genome organizationand regulation. However, identifying and analyzing thesestructural features require advanced computational methods. In recent years, machine learning approaches, particularly deeplearning, have emerged as powerful tools for detecting andanalyzing structural information. In this review, we present anoverview of various machine learning-based techniques for determiningchromosomal organization. Starting with the progress inpredicting interactions from DNA sequences, we describe methodsfor identifying various hierarchical structures from Hi-C data. Additionally, we present advances in enhancing the chromosomecontact frequency map resolution to overcome the limitationsof Hi-C data. Finally, we identify the remaining challenges andpropose potential solutions and future directions.
Abstract
The study of chromatin interactions has advanced considerablywith technologies such as high-throughput chromosome conformationcapture (Hi-C) sequencing, providing a genome-wide viewof physical interactions within the nucleus. These techniqueshave revealed the existence of hierarchical chromatin structuressuch as compartments, topologically associating domains (TADs),and chromatin loops, which are crucial in genome organizationand regulation. However, identifying and analyzing thesestructural features require advanced computational methods. In recent years, machine learning approaches, particularly deeplearning, have emerged as powerful tools for detecting andanalyzing structural information. In this review, we present anoverview of various machine learning-based techniques for determiningchromosomal organization. Starting with the progress inpredicting interactions from DNA sequences, we describe methodsfor identifying various hierarchical structures from Hi-C data. Additionally, we present advances in enhancing the chromosomecontact frequency map resolution to overcome the limitationsof Hi-C data. Finally, we identify the remaining challenges andpropose potential solutions and future directions.