Rational Design of Profile Hidden Markov Models for Viral Classification and Discovery

Abstract

This chapter provides an overview of the theoretical concepts and practical applications of methods for the rational design and application of profile hidden Markov models (profile HMMs) in viral discovery and classification. Profile HMMs are probabilistic models that represent sequence diversity and constitute a very sensitive approach for detecting remote homologs. One of the most relevant and challenging applications of profile HMMs is the discovery of viruses in metagenomic samples, a fundamental task for epidemiological surveillance. In this chapter, publicly available resources of viral profile HMMs are presented and the methods involved in their construction are discussed. Several aspects to be considered for the generation of profile HMMs are presented, including technical pitfalls that should be avoided, and the potential applications of such models for detecting specific viral sequences. This chapter also introduces a bioinformatics application that implements methods to select informative regions of a multiple sequence alignment and build profile HMMs with different taxonomic specificities. Additional programs using profile HMMs for targeted sequence assembly and detection of multigene entities are also presented. Such programs, integrated into a common framework for viral research,

Rational Design of Profile HMMs for Sensitive and Specific Sequence Detection with Case Studies Applied to Viruses, Bacteriophages, and Casposons

“…The tag (e.g., CUTOFF SCORE 45) itself is neutral to programs of the HMMER package (hmmsearch and nhmmer) but can be specified by using the parameter -T of these programs (e.g., -T 45). Some bioinformatics applications developed by our group, such as GenSeed-HMM [ 47 ], HMM-Prospector ( —URL accessed on 2 February 2023) [ 52 ], and e-Finder ( —URL accessed on 2 February 2023) [ 52 ] can automatically read this tag in each profile HMM and use its respective value in the parameter -T when invoking hmmsearch for the similarity searches.…”

“…All these tools can automatically use cutoff scores assigned by TABAJARA. A recent review [ 52 ] describes an integrated approach for profile HMM construction in the possible applications of these models.…”

“…Profile HMMs have been increasingly used in virus studies, including astrovirus [ 44 ], Influenza A viruses [ 45 , 46 ], Microviridae phages [ 47 ], coronaviruses [ 48 ], mitovirus [ 49 ], herpesviruses [ 49 , 50 ], and African swine fever virus (ASFV) [ 51 ]. Additionally, profile HMMs can be used as seeds to nucleate sequencing reads in progressive assemblies for a targeted specific reconstruction of viral sequences from metagenomic datasets [ 34 , 47 , 52 ]. Several different databases of viral profile HMMs have been developed and are publicly available, such as pVOGs [ 53 ] for prokaryotic viruses and vFam for eukaryotic viruses [ 54 ].…”

Mining livestock genome datasets for an unconventional characterization of animal DNA viromes

“…Alternative sequence mining approaches (e.g. profile Hidden Markov Models [90]) could be potentially useful to detect some of these missed viruses. A few other biases, which might affect the general interpretation of the obtained results, could be also derived from the way in which we set the bioinformatic pipeline and from intrinsic characteristics of the mined WGS datasets deposited in the public repository.…”

Mapping Data to Deep Understanding: Making the Most of the Deluge of SARS-CoV-2 Genome Sequences

“…Another approach is to build profile hidden Markov models (HMMs), which can identify taxonomic lineages and variants of viruses. HMMs have been used to align SARS-CoV-2 sequences and compare its spike protein to that of other coronaviruses ( 22 , 54 , 55 ).…”