Weizhong Li received his Ph.D. in Physical Chemistry from Nankai University in 1996 with Prof. Fangming Miao. Between 1996 and 1999, he worked in Beijing University with Prof. Luhua Lai on bioinformatics and computational chemistry. He started Postdoctoral research in 1999 with Dr. Adam Godzik at San Diego Supercomputer Center and then the Burnham Institute focusing on development of bioinformatics algorithm, software, and web-servers. In 2002, he moved to Quorex Inc, a pharmaceutical company later acquired by Pfizer, where he led the company's bioinformatics effort in drug discovery. He join the Burnham Institute as staff scientist in 2004 and then UCSD as senior scientist in 2006 working on structural genomics, functional proteomics and metagenomics.

Weizhong Li's current research interests include method development for metagenomics, large scale protein annotation, gene finding, sequence clustering, protein family classification, structural bioinformatics, and cheminformatics. The bioinformatics software he developed are highly cited and widely used in many labs and institutions such as Uniprot, PDB and EBI.

Tentative Title

Probing Metagenomics by Rapid Analysis of Mega-datasets

Abstract

The field of metagenomics adds a new path to explore the great diversity of microbial world, evidenced by tens of millions of newly discovered metagenomic sequences. In the meantime, it creates new challenges in data analysis, methodologically and computationally. The gigantic size of metagenomic data prevents many conventional computational analyses and makes it impossible for some large-scale investigations. Here, we introduce an efficient pipeline, called RAMMCAP, involving ultra-fast clustering, comparing and annotation. It includes a novel approach and a unique visualization interface to compare metagenomes based on cluster analysis and annotation. We also propose new function-centric (as opposed to organism-centric) approaches to study environmental microbes. Two largest available metagenomic collections, the ¡°Global Ocean Sampling¡± and the metagenomic profiling of ¡°Nine Biomes¡±, with 7.7 and 14.6 million reads respectively, were studied with our method. With just moderate compute efforts, we can quickly analyze these extremely large metagenomic datasets, providing rich information from a global view of data to details of specific aspects within or between datasets including gene family distribution, novel genes, annotation, sample comparison, population and taxonomy, and so on. We made systematic functional comparison for metagenomic samples and identified significant patterns of functional similarities and variances. Our findings are available online from http://tools.camera.calit2.net/camera/rammcap.