From promoter sequence to expression: a probabilistic framework
Proceedings of the sixth annual international conference on Computational biology
A Simple Hyper-Geometric Approach for Discovering Putative Transcription Factor Binding Sites
WABI '01 Proceedings of the First International Workshop on Algorithms in Bioinformatics
Sparse bayesian learning and the relevance vector machine
The Journal of Machine Learning Research
DWE: Discriminating Word Enumerator
Bioinformatics
Statistics of local multiple alignments
Bioinformatics
A boosting approach for motif modeling using ChIP-chip data
Bioinformatics
| Hi-index | 0.00 |
Technological advances in experimental and computational molecular biology have revolutionized the whole fields of biology and medicine. Large-scale sequencing, expression and localization data have provided us with a great opportunity to study biology at the system level. I will introduce some outstanding problems in genome expression and regulation network in which better modern statistical and machine learning technologies are desperately needed. Recent revolution in genomics has transformed life science. For the first time in history, mankind has been able to sequence the entire human own genome. Bioinformatics, especially computational molecular biology, has played a vital role in extracting knowledge from vast amount of information generated by the high throughput genomics technologies. Today, I am very happy to deliver this key lecture at the First International Conference on Pattern Recognition and Machine Intelligence at the world renowned Indian Statistical Institute (ISI) where such luminaries as Mahalanobis, Bose, Rao and others had worked before. And it is very timely that genomics has attracted new generation of talented young statisticians, reminding us the fact that statistics was essentially conceived from and continuously nurtured by biological problems. Pattern/rule recognition is at the heart of all learning process and hence of all disciplines of sciences, and comparison is the fundamental method: it is the similarities that allow inferring common rules; and it is the differences that allow deriving new rules. Gene expression, normally referring to the cellular processes that lead to protein production, is controlled and regulated at multiple levels. Cells use this elaborate system of “circuits” and “switches” to decide when, where and by how much each gene should be turned on (activated, expressed) or off (repressed, silenced) in response to environmental clues. Genome expression and regulation refer to coordinated expression and regulation of many genes at large-scales for which advanced computational methods become indispensable. Due to space limitations, I can only highlight some of the pattern recognition problems in transcriptional regulation, which is the most important and best studied. Currently, there are two general outstanding problems in transcriptional regulation studies: (1) How to find the regulatory regions, in particular, the promoters regions in the genome (throughout most of this lecture, we use promoter to refer to proximal promoters, e.g. ~ 1kb DNA at the beginning of each gene); (2) How to identify functional cis-regulatory DNA elements within each such region.