MacSyFinder v2: Improved modelling and search engine to identify molecular systems in genomes

Abstract

Complex cellular functions are usually encoded by a set of genes in one or a few organized genetic loci in microbial genomes. MacSyFinder uses these properties to model and then annotate cellular functions in microbial genomes. This is done by integrating the identification of each individual gene at the level of the molecular system. We hereby present a major release of MacSyFinder (Macromolecular System Finder), MacSyFinder version 2 (v2). This new version is coded in Python 3 (>= 3.7). The code was improved and rationalized to facilitate future maintainability. Several new features were added to allow more flexible modelling of the systems. We introduce a more intuitive and comprehensive search engine to identify all the best candidate systems and sub-optimal ones that respect the models' constraints. We also introduce the novel macsydata companion tool that enables the easy installation and broad distribution of the models developed for MacSyFinder (macsy-models) from GitHub repositories. Finally, we have updated, improved, and made available MacSyFinder popular models for this novel version: TXSScan to identify protein secretion systems, TFFscan to identify type IV filaments, CONJscan to identify conjugative systems, and CasFinder to identify CRISPR associated proteins.

Identification of protein secretion systems in bacterial genomes using MacSyFinder version 2

“…TXSScan is a set of pre-defined MacSyFinder models and profiles (also called “macsy-model” package), to detect the best-studied protein secretion systems in diderms ( see Figure 1 , with the T1SS, T2SS, T3SS, T4SS, T5SS, T6SS, T9SS) and related appendages with homologs in diderm and monoderm bacteria, and even in archaea ( see Figure 2 , with the bacterial flagellum and the pili members of the type IV filament super-family) [10, 12, 19]. It has to be noted that since MacSyFinder version 2, it is possible to class MacSyFinder models in sub-folders with a hierarchy enabling to run MacSyFinder only on a subset of relevant models at once.…”

“…This information can be put together in a model of the system that can be used by a computer program that we developed and called MacSyFinder for “Macromolecular System Finder” [18]. This program, now in its second version (“v2”), searches genomes for instances satisfying the characteristics described in the model in terms of genetic content (or “quorum”) and organization [18, 19].…”

“…MacSyFinder uses Hidden Markov model (HMM) protein profiles specified in a model to search for components of a system in a file of protein sequences. In a nutshell, it collects clusters of co-localized components in the genome and checks if combinations of these clusters may result in valid systems relative to the gene content and co-localization specified in the systems’ models [19]. MacSyFinder then outputs the results of the protein profile searches and the information about the identified secretion systems.…”

Identification and characterization of thousands of bacteriophage satellites across bacteria

“…Instead, we made a curated annotation of a large set of known satellite elements and used them to iteratively identify core genes of each family of satellites. We then designed individual and customizable MacSyFinder (22, 23) models that represent the genetic composition of the different phage satellite families ( Fig S1 ). The models were used in MacSyFinder to identify occurrences of the putative satellites.…”

“…When this was not the case, we built custom HMM profiles by aligning all sequences of the family with Clustal Omega (26) (Version 1.2.3) with default parameters, and then by using hmmbuild (default parameters) from hmmer 3.3.2 (27). We used MacSyFinder (22,23) (version 2.05rc) to provide a reproducible, shareable, and easily modifiable tool to identify phage satellites in bacterial genomes. Briefly, MacSyFinder searches for co-occurrence of the markers of each phage satellite family in bacterial genomes.…”