Xiaowo Wang
Title：De novo design of gene regulatory codes using AI ABSTRACT
Abstract: DNA is the programming language of life. Gene expression level is determined by the regulatory information encoded in DNA sequence elements like promoters. Obtaining new genetic regulatory elements has tremendous usage in metabolic engineering and synthetic biology applications. Theoretically, there are as many as 4100 possible combinations even for a 100 base pair DNA sequence. Only part of them are biocompatible and naturally occurring genomes make up a very small subset. To explore the high dimensional space of potential sequences, we reported a novel machine learning framework for de novo gene regulatory sequence design. The model, which was guided by sequence features learned from natural DNA sequences, could capture long range dependencies between nucleotides at different positions and design novel synthetic elements in silico. The model designed gene promoters were experimentally demonstrated to be functional in vivo, and a number of them showed comparable or even higher activities than most active natural promoters and their strongest mutants. Many of these generated sequences showed low global sequence similarity to the wild type genome, and noncanonical motifs were found in highly expressed promoters. Our work provided new insights into de novo gene regulatory element design, indicating the potential ability of AI to obtain new optimized genetic elements.
Biography: Dr. Xiaowo Wang is currently a full professor at the Department of Automation Tsinghua University. He received his bachelor's degree of engineering and Ph.D. in bioinformatics from Tsinghua University. He was a visiting student in Cold Spring Harbor laboratory in 2007, and a Tang Distinguished Scholar in Quantitative Biology Institute QB3 of UC Berkeley in 2012. He joined the faculty of Tsinghua University since 2008. His lab aims to bring machine learning and biology approaches together to understand gene regulation networks systematically, and guide the quantitative design of synthetic biological systems for precise medicine applications. He has published 50+ peer-reviewed papers in journals including PNAS, Cell reports, NAR, Bioinformatics etc., and his work has been cited more than 7,000 times. He is a receipt of the National Natural Science Fund for Excellent Young Scholars of China in 2013, and Young Scientists Awards of Chinese Association of Automation in 2019.
Abstract: DNA is the programming language of life. Gene expression level is determined by the regulatory information encoded in DNA sequence elements like promoters. Obtaining new genetic regulatory elements has tremendous usage in metabolic engineering and synthetic biology applications. Theoretically, there are as many as 4100 possible combinations even for a 100 base pair DNA sequence. Only part of them are biocompatible and naturally occurring genomes make up a very small subset. To explore the high dimensional space of potential sequences, we reported a novel machine learning framework for de novo gene regulatory sequence design. The model, which was guided by sequence features learned from natural DNA sequences, could capture long range dependencies between nucleotides at different positions and design novel synthetic elements in silico. The model designed gene promoters were experimentally demonstrated to be functional in vivo, and a number of them showed comparable or even higher activities than most active natural promoters and their strongest mutants. Many of these generated sequences showed low global sequence similarity to the wild type genome, and noncanonical motifs were found in highly expressed promoters. Our work provided new insights into de novo gene regulatory element design, indicating the potential ability of AI to obtain new optimized genetic elements.
Biography: Dr. Xiaowo Wang is currently a full professor at the Department of Automation Tsinghua University. He received his bachelor's degree of engineering and Ph.D. in bioinformatics from Tsinghua University. He was a visiting student in Cold Spring Harbor laboratory in 2007, and a Tang Distinguished Scholar in Quantitative Biology Institute QB3 of UC Berkeley in 2012. He joined the faculty of Tsinghua University since 2008. His lab aims to bring machine learning and biology approaches together to understand gene regulation networks systematically, and guide the quantitative design of synthetic biological systems for precise medicine applications. He has published 50+ peer-reviewed papers in journals including PNAS, Cell reports, NAR, Bioinformatics etc., and his work has been cited more than 7,000 times. He is a receipt of the National Natural Science Fund for Excellent Young Scholars of China in 2013, and Young Scientists Awards of Chinese Association of Automation in 2019.