 |
|
报告题目: Modern biostatistical techniques and machine learning methods in Predicting Long Non-Coding RNAs: Recent Progress, Challenges, Prospects
时间: 1月6号下午3点
地点: 8号楼学院小报告厅
报告人: Tianhua (Tim) Niu, MS, ScD
Dr. Niu is currently Clinical Assistant Professor in Department of Global Biostatistics and Data Science at Tulane University School of Public Health and Tropical Medicine. Dr.Niu’s research interests include Population Genetics, Computational Statistics, Transcriptomics, Bioinformatics, and has published over 110 papers in international peer-reviewed journals.
Abstract: Although < 2% of the human genome DNA comprises protein-coding genes, the majority of the human genome is transcribed as non-coding RNA transcripts. Long Non-Coding RNAs (lncRNAs), endogenous non-coding RNAs of more than 200 nucleotides in length, have important biological functions. The role of lncRNAs in mediating disease pathogenesis is increasingly being elucidated by findings from genome-wide association studies as well as transcriptome-based studies. E.g., experimental studies have increasingly shown that lncRNAs regulate key pathways in many cancer types (e.g., breast cancer, colorectal cancer, and hepatocellular carcinoma). Modern Biostatistical [e.g., least absolute shrinkage and selection operator (LASSO) (Tibshirani, 1996) and elastic net (Zou and Hastie, 2005)] and machine learning [e.g., various clustering algorithms, support vector machine (SVM), logistic regression model, and Bayesian network] methods, which include both supervised and unsupervised techniques, are widely used in biostatistics and computational and systems biology. Based on several major machine learning algorithms (particularly SVM), a set of bioinformatics tools, e.g., CPAT, CPC, CONC, and others, have been developed for predicting lncRNAs. A synopsis of modern biostatistical techniques and machine learning methods is provided, and several strengths and limitations of existing lncRNA prediction tools are illustrated using specific examples. By applying effective biostatistical algorithms and machine learning methodologies built on solid mathematical foundations, the tremendous potential of discoveries of lncRNAs as biomarkers for novel diagnosis and therapeutic targets for human diseases is illustrated.
|
|
|
