From DNA Sequences to Gene Expression Modeling - از دنباله هاى دی‌ان‌آ تا مدل‌سازی بیان ژن

توضیحات:

ارائه فرزانه خواجویی در
سمینار زمستانی مباحث پیشرفته در علوم و مهندسی کامپیوتر
Talk By Farzaneh Khajouei
Talk Title: Interpreting non-coding SNPs and quantifying the value of perturbation experiments using ensembles of biophysical models

Abstract: The ability to predict phenotype solely from DNA sequences is a powerful tool for biological and medical research. When models that give such predictions are reasonably accurate, minor genetic variations – even a single base pair – can predict important functional impact, such as problems in embryo development. This project seeks to consider ‘enhancer’ sequences when doing such modeling. Enhancers, regions of DNA of around 1000 base pairs, provide an entry point for the cellular mechanisms responsible for reading DNA and producing the corresponding proteins. In recent years, several powerful quantitative models that map a given enhancer sequence to its expression readout in varying cellular conditions has been developed. The success of these models, especially those based on equilibrium thermodynamics, has been demonstrated in the context of systems such as gene regulation in early-stage Drosophila embryos.

We are using GEMSTAT (a thermodynamic-based method previously developed in our research group) to computationally predict the gene expressions from enhancer sequences of Drosophila Melanogaster while considering the uncertainty in the parameters of the model that GEMSTAT uses. Modeling gene expression from enhancer sequences often involves training several parameters. In high dimensional parameter space, model-training algorithms usually converge to local optima, where different parameter setups result in models that explain the data equally well. Our goal is to characterize the properties of a parameter space with a probability distribution over well fitting parameter sets, in order to take the uncertainty of the parameter space into account for sequence-to-expression modeling. We are adopting this framework to further improve our understanding of the parameter space, while characterizing features of the modeling procedure. This gives us a systematic way to incorporate the modeling uncertainty into our predictions, making our predictions more robust.
دانشکده کامپیوتر دانشگاه صنعتی شریف
انجمن علمی دانشکده کامپیوتر دانشگاه صنعتی شریف
Winter Seminar Series - WSS2015
Advanced Topics in Computer Science and Engineering
Sharif University of Technology
Students Scientific Chapter
ssc.ce.sharif.edu/wss2015
https://youtu.be/iPtNM76Go3k

لغات کلیدی:

سمینار, زمستانی, علوم, کامپیوتر, دانشگاه, شریف, Sharif, University, of, Technology, Bioinformatics, DNA, Sequences, WSS

درباره ما | تماس با ما | قوانین تخته سفید