Research work · 2025
Advanced Probability for Bioinformatics Applications
Concepts in Bioinformatics and Genomics
Abstract
This chapter introduces advanced probabilistic models essential for evaluating sequence alignments in bioinformatics. It begins by explaining how extreme value distribution quantify the significance of sequence similarities, focusing on the interpretation of P- and E-values derived from Basic Local Alignment Search Tool (BLAST) scores. It then explores Markov chains and hidden Markov models (HMMs), detailing how state transitions and emissions help infer biological patterns, such as CpG islands or conserved motifs. Using training sequences, the chapter demonstrates how to initialise, train, and optimise profile HMMs for multiple sequence alignment, employing algorithms like Baum–Welch and Viterbi for parameter estimation and path identification. Finally, the chapter underscores how these models predict relatedness and functional similarity across sequences with statistical rigour.