Accuratede novoidentification of biosynthetic gene clusters with GECCO

Abstract

Biosynthetic gene clusters (BGCs) are enticing targets for (meta)genomic mining efforts, as they may encode novel, specialized metabolites with potential uses in medicine and biotechnology. Here, we describe GECCO (GEne Cluster prediction with COnditional random fields; https://gecco.embl.de), a high-precision, scalable method for identifying novel BGCs in (meta)genomic data using conditional random fields (CRFs). Based on an extensive evaluation of de novo BGC prediction, we found GECCO to be more accurate and over 3x faster than a state-of-the-art deep learning approach. When applied to over 12,000 genomes, GECCO identified nearly twice as many BGCs compared to a rule-based approach, while achieving higher accuracy than other machine learning approaches. Introspection of the GECCO CRF revealed that its predictions rely on protein domains with both known and novel associations to secondary metabolism. The method developed here represents a scalable, interpretable machine learning approach, which can identify BGCs de novo with high precision.

Evolutionary investigations of the biosynthetic diversity in the skin microbiome usinglsaBGC

“…Further, an improved understanding of the relationship between intra-taxa genomic diversity and BGC content novelty at the baseresolution can also highlight which sub-lineages or sub-clades bear the largest reservoir of untapped secondary metabolic potential. Finally, we posit that identifying evolutionary trends of BGCs detected by highly-reliable rule-based approaches, such as antiSMASH 8 , can be used to assess the validity of BGC predictions from newly developed machine learning approaches, such as DeepBGC 11 or GECCO 65 , which we now support usage of in lsaBGC.…”

Pyrodigal: Python bindings and interface to Prodigal, an efficient method for gene prediction in prokaryotes

“…Pyrodigal has already been used as the implementation for the initial ORF finding stage in several domains, including biosynthetic gene cluster prediction (Carroll et al, 2021), prophage identification (Sirén et al, 2021;Turkington et al, 2021), and pangenome analysis (Hernández et al, 2021).…”

Genome Mining as an Alternative Way for Screening the Marine Organisms for Their Potential to Produce UV-Absorbing Mycosporine-like Amino Acid

“…Analyses of genomic sequences provided a hidden glance into organisms’ potential for producing bioactive NPs. Modern strategies in genome mining include in silico methods for BGC identification to facilitate the discovery of novel NPs [ 91 ]. For genome mining, different types of sequence data (e.g., genomics, transcriptomics, metabolomics, proteomics, epigenomics, and multi-omics) [ 10 , 40 , 54 ] and numerous bioinformatics tools are applied [ 67 , 92 ].…”

“…Genome mining search for novel antibiotics, Antibiotic-Resistant Target Seeker (ARTS), is available at (accessed on 12 June 2022) [ 121 ]. Recently, a new machine learning approach, GECCO (GEne Cluster prediction with COnditional random fields; ; accessed on 12 June 2022), allowed much higher identification of de novo BGCs from metagenomics data, confirming the important link for prediction between protein domain and secondary metabolites [ 91 ]. An overview of different bioinformatics tools applied for genome mining and useful in the discovery of BCGs and related secondary metabolites are presented in Table 2 .…”