|
Eric Lu Zhang Assistant Professor Department of Computer Science Hong Kong Baptist University Kowloon Tong, Hong Kong ericluzhang AT comp DOT hkbu DOT edu DOT hk Phone: (852) 3411-5880 Fax: (852) 3411-7892 Office: Room R708, Sir Run Run Shaw Building |
For a complete list of my publications and other detailed info, please see my Google Scholar page or CV.
Openings: I am looking for talented and self-motivated PhD students, Postdocs and Research Assistant Professors who are interested in computational genomics and machine learning in genomics. Please contact me by email with your CV and research proposal.
The human genome holds the key for understanding the genetic basis of human evolution, hereditary illnesses and many phenotypes. Whole-genome reconstruction and variant discovery, accomplished by analysis of data from whole-genome sequencing experiments, are foundational for the study of human genomic variation and analysis of genotype-phenotype relationships. Over the past decades, cost-effective whole-genome sequencing has been revolutionized by short-fragment approaches, the most widespread of which have been the consistently improving generations of the original Solexa technology, now referred to as Illumina sequencing. An alternative approach is offered by the 10x Genomics Chromium system and stLFR from BGI, which distributes the DNA prep into millions of partitions or beads where specific barcode sequences are attached to short amplification products that are templated off the input fragments. But there lacks efficient software to handle this recently emergent technology and make full use of DNA long-range information. We aim to develop a series of computational tools to analyze linked-reads data, including read alignment, de novo assembly, variant detection, evolutionary analysis et al. We believe our contribution can move us one step further to make precision medicine into reality.
Besides human genome analysis, we also design algorithms to decipher metagenome from linked-read and long-read sequencing.
A key public health challenge is to identify individuals at high-risk for common diseases in order to enable prescreening or preventive therapies. Much effort has been made in identifying disease causal genomic variants and evaluating their contribution in disease prediction. Unlike single gene diseases that are usually caused by inherited monogenic mutations, common diseases have multifactorial etiologies that involve the interplay of both genetic and non-genetic factors. Therefore, how to effectively identify high-risk incident cases from "multi-level" information are core goals for precision medicine. We aims to develop machine learning (especially deep learning) algorithms on integrating genomic data, clinical images, clinical records and lifestyle information to predict human complex diseases.
The missing heritability of human complex diseases is a critical problem in biomedical research. Traditional statistical approaches are unable to make full use of the accumulated genomic big data and may miss complex, nonlinear relationships of risk factors. We aim to develop novel deep learning methods by integrating trans-omics big data to explore the "dark regions" in genomic studies. Single-cell RNA sequencing is an emerging technology which provides us the opportunity to observe the gene expression profiles at single-cell resolution. Analysing scRNA-seq data can help us understand the heterogeneity of different cell types and capture cell differentiation naturally. We aim to design and apply modern machine learning approaches to address a series of computational problems in scRNA-seq.