HMMerge

HMMerge aligns multiple sequences using a merged ensemble profile Hidden Markov Model to improve multiple sequence alignment accuracy for datasets with substantial sequence length heterogeneity, including incorporation of very short sequences.

Key Features:

• Handling Length Heterogeneity: Targets datasets with substantial variation in sequence lengths to reduce alignment errors caused by length heterogeneity.
• Integration of Short Sequences: Incorporates very short query sequences into an existing backbone alignment.
• Ensemble Profile Hidden Markov Models: Represents the backbone alignment as an ensemble of profile HMMs, building on methodologies from UPP and WITCH.
• Merged HMM Alignment: Constructs a merged HMM from the ensemble and uses the merged HMM to align query sequences.

Scientific Applications:

• Phylogenetic inference: Produces improved MSAs for downstream phylogenetic studies.
• Functional annotation: Enhances alignment of homologous regions across length-heterogeneous datasets to support functional annotation.
• Evolutionary analysis: Supports analyses involving sequences affected by large deletions or extreme length variation.

Methodology:

Represents a backbone alignment as an ensemble of profile HMMs (building on UPP and WITCH), constructs a merged HMM from that ensemble, and uses the merged HMM to align query sequences including very short sequences.