學(xué)術(shù)沙龍主題: PB-DiffHiC: a statistically principled method for detecting differential chromatin interactions using raw pseudo-bulk Hi-C data
報(bào)告人: 周彥，深圳大學(xué)教授
報(bào)告時(shí)間: 2025年4月19日(周六)；下午3:30—5:00
報(bào)告地點(diǎn)：南校區(qū)網(wǎng)絡(luò)安全大樓 121 會(huì)議室
邀請(qǐng)人：李本崇
報(bào)告人簡(jiǎn)介：周彥，深圳大學(xué)數(shù)學(xué)科學(xué)學(xué)院教授，博士生導(dǎo)師，統(tǒng)計(jì)與數(shù)據(jù)科學(xué)系主任。畢業(yè)于東北師范大學(xué)，曾在UIUC從事博士后工作，2015年進(jìn)入深圳大學(xué)工作。曾訪問(wèn)香港大學(xué)，香港浸會(huì)大學(xué)等。主要研究方向?yàn)樯锝y(tǒng)計(jì)，機(jī)器學(xué)習(xí)，醫(yī)學(xué)統(tǒng)計(jì)等。獲得深圳市孔雀計(jì)劃獎(jiǎng)勵(lì)C類。主持國(guó)家面上項(xiàng)目，國(guó)家青年項(xiàng)目等數(shù)項(xiàng)。以第一或通訊作者身份在Genome Research（影響因子：14.38），Bioinformatics（影響因子：7.38），Statistics in Medicine, BMC Genomics等期刊上發(fā)表SCI論文四十余篇。兼職廣東省高等學(xué)校教學(xué)指導(dǎo)委員會(huì)委員；廣東省現(xiàn)場(chǎng)統(tǒng)計(jì)協(xié)會(huì)副理事長(zhǎng)，常務(wù)理事；中國(guó)現(xiàn)場(chǎng)統(tǒng)計(jì)協(xié)會(huì)理事；中國(guó)環(huán)境資源統(tǒng)計(jì)學(xué)會(huì)和多元統(tǒng)計(jì)學(xué)會(huì)常務(wù)理事等。
報(bào)告摘要: Single-cell Hi-C data provide unprecedented opportunities for analyzing differential chromatin interactions, essential for understanding genome structure-function relationships across various biological conditions. However, differential chromatin interaction analysis requires Hi-C data at high resolutions (e.g., 10 Kb), and due to the sparsity of scHi-C data, existing methods typically rely on single cell imputation or conventional bulk approaches, which can comprise the reliability of differential interaction detection. Here, we present PB-DiffHiC, a new optimized parametric statistical framework that directly analyzes the raw pseudo-bulk Hi-C data at 10 Kb resolution between conditions. PB-DiffHiC incorporates Gaussian convolution, stability of short-range interactions, and Poisson distribution to enable joint normalization and detection of significant differential chromatin interactions. Benchmark using cell-type-specific chromatin loops shows that PB-DiffHiC achieves higher precision than alternative methods. Applying PB-DiffHiC to pseudo-bulk and matched bulk Hi-C data demonstrate stronger concordance with identified differential interactions, reinforcing reliability. In a case study, PB-DiffHiC successfully identifies Kcnq5-associated differential interactions, closely matching SnapHiC-D results despite not relying on single-cell imputation. Therefore, PB-DiffHiC is a statistically sound method for enhanced comparative analysis of chromatin interactions using raw pseudo-bulk Hi-C data at 10 Kb resolution. The source code of PB-DiffHiC is publicly available at https://github.com/Tian-Dechao/PB-DiffHiC.
