Bayesian model-based approaches with MCMC computation to some bioinformatics problems

摘要

Bioinformatics applications can address the transfer of information at several stagesof the central dogma of molecular biology, including transcription and translation.This dissertation focuses on using Bayesian models to interpret biological data inbioinformatics, using Markov chain Monte Carlo (MCMC) for the inference method.First, we use our approach to interpret data at the transcription level. We proposea two-level hierarchical Bayesian model for variable selection on cDNA Microarraydata. cDNA Microarray quantifies mRNA levels of a gene simultaneously so hasthousands of genes in one sample. By observing the expression patterns of genes undervarious treatment conditions, important clues about gene function can be obtained.We consider a multivariate Bayesian regression model and assign priors that favorsparseness in terms of number of variables (genes) used. We introduce the use ofdifferent priors to promote different degrees of sparseness using a unified two-levelhierarchical Bayesian model. Second, we apply our method to a problem related tothe translation level. We develop hidden Markov models to model linker/non-linkersequence regions in a protein sequence. We use a linker index to exploit differencesin amino acid composition between regions from sequence information alone. A goalof protein structure prediction is to take an amino acid sequence (represented asa sequence of letters) and predict its tertiary structure. The identification of linkerregions in a protein sequence is valuable in predicting the three-dimensional structure.Because of the complexities of both models encountered in practice, we employ theMarkov chain Monte Carlo method (MCMC), particularly Gibbs sampling (Gelfandand Smith, 1990) for the inference of the parameter estimation.