COMET: adaptive context-based modeling for ultrafast HIV-1 subtype identification

Abstract

Viral sequence classification has wide applications in clinical, epidemiological, structural and functional categorization studies. Most existing approaches rely on an initial alignment step followed by classification based on phylogenetic or statistical algorithms. Here we present an ultrafast alignment-free subtyping tool for human immunodeficiency virus type one (HIV-1) adapted from Prediction by Partial Matching compression. This tool, named COMET, was compared to the widely used phylogeny-based REGA and SCUEAL tools using synthetic and clinical HIV data sets (1 090 698 and 10 625 sequences, respectively). COMET's sensitivity and specificity were comparable to or higher than the two other subtyping tools on both data sets for known subtypes. COMET also excelled in detecting and identifying new recombinant forms, a frequent feature of the HIV epidemic. Runtime comparisons showed that COMET was almost as fast as USEARCH. This study demonstrates the advantages of alignment-free classification of viral sequences, which feature high rates of variation, recombination and insertions/deletions. COMET is free to use via an online interface.

Comparative Evaluation of Subtyping Tools for Surveillance of Newly Emerging HIV-1 Strains

“…On one hand, Stanford-v8.1 showed excellent performance in sensitivity (97.4% to 100%) on other pure subtypes (D, F1, and F2) in contrast to other rapid tools (reporting Ͻ90% sensitivity); this indicates that Stanford-v8.1 might be acceptable in assigning D or F strains. On the other hand, only COMET or REGA-v3 might be acceptable in assigning viruses as subtype A1 or G, respectively (27,29). Regarding the sensitivity to recombinant forms, Stanford-v8.1 appeared highly reliable in assigning CRF01_AE, CRF02_AG, CRF06_cpx, CRF11_cpx, and CRF36_cpx.…”

“…These tools have different characteristics in inferring subtypes (26). In particular, because REGA-v3 and SCUEAL are phylogenetics-based tools, their computation times are longer than those of other tools that are based on a statistical (COMET) or similarity (Stanford HIVdb) approach (24,26,28,29).…”

The International Dimension of the U.S. HIV Transmission Network and Onward Transmission of HIV Recently Imported into the United States

“…Subtype classification was performed using COMET. 27 Non-B clusters with multiple potential transmission partners within the United States were subjected to molecular clock analysis in a Bayesian Markov-chain Monte Carlo framework in BEAST v1.8.0. 28,29 Each molecular clock analysis was run for 10,000,000 generations, sampling every 2,500 generations.…”

Exploring resistance pathways for first-generation NS3/4A protease inhibitors boceprevir and telaprevir using Bayesian network learning

“…HCV genotypes and subtypes were assigned using the Oxford HCV automated subtyping tool version 2 (Alcantara et al, 2009;De Oliveira et al, 2005), COMET (Struck et al, 2014) and by manual phylogenetic analysis using reference sequences from the Los Alamos HCV sequence database (Kuiken et al, 2008). Sequences were aligned against reference sequence H77 (NC_004102) using an in-house developed codon aware alignment tool (Cuypers et al, 2015).…”