show_emfit

show_emfit applies Hidden Markov Models (HMMs) and the Expectation-Maximization (EM) algorithm to fit model parameters and segment biological sequences by reconstructing hidden state paths with the forward-backward algorithm.

Key Features:

• Model fitting with EM algorithm: Employs the Expectation-Maximization (EM) algorithm to fit Hidden Markov Models (HMMs) to biological sequence data and optimize model parameters to maximize data likelihood.
• Sequence segmentation via forward-backward algorithm: Uses the forward-backward algorithm to infer the most probable hidden state path and segment sequences into constituent states.
• Flexibility with fixed parameters: Supports analyses using fixed, user-provided model parameters to perform segmentation without re-optimization.

Scientific Applications:

• Genomic analysis: Applicable to genome-scale analyses such as the Lactobacillus sakei strain 23K sequencing project for identifying genes involved in purine nucleoside scavenging and responses to changing redox conditions.
• Microbial ecology and biopreservation: Facilitates identification of genes associated with biofilm formation and cellular aggregation relevant to microbial colonization and biopreservation in fermented meats.

Methodology:

Implements Hidden Markov Models with Expectation-Maximization for parameter learning and the forward-backward algorithm for hidden-state inference, and can operate with fixed model parameters for segmentation-only analyses.