Can Random Walking on a Hi-C Contact Matrix Lead to Data Quality Improvement? An Assessment

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• High-throughput chromosome conformation capture (Hi-C) and single-cell Hi-C (scHi-C) are crucial for studying genome-wide chromatin organization, including compartments, TADs, and long-range interactions.
• Data quality, particularly sparsity in scHi-C, necessitates improvement methods before biological analyses.
• Random walk-based methods, including RWS and RWR, have been developed for data quality enhancement but lack rigorous justification and performance evaluation.

Purpose of the Study:

• To characterize random-walk-based approaches for chromatin organization analysis.
• To empirically investigate the performance of random walks in identifying Topologically Associated Domains (TADs) before and after applying these methods.
• To provide guidance on parameter selection for random walk methods.

Main Methods:

• Analysis of random-walk-based methods (RWS and RWR) characteristics.
• Empirical investigation of TAD identification performance.
• Extensive analytical analyses, simulation studies, and real data applications.

Main Results:

• Random walk methods, even with optimal parameter tuning, showed minimal improvement in identifying TADs.
• Little to no enhancement in downstream analyses was observed after applying random walk methods.
• Lack of practical guidelines for parameter selection (e.g., number of steps for RWS, restart probability for RWR) hinders effective application.

Conclusions:

• The study advises caution regarding the use of random-walk-based approaches for Hi-C and scHi-C data quality improvement.
• Random walk methods do not consistently enhance the identification of crucial genomic structures like TADs.
• Researchers should carefully consider the limited benefits and lack of clear parameter guidelines before employing these methods in their analyses.