bioRxivBioinformaticsPreprint
MetaGeneMark-2: Improved Gene Prediction in Metagenomes
Abstract
Accurate prediction of protein-coding genes in metagenomic contigs presents a well-known challenge. Particularly difficult is to identify short and incomplete genes as well as positions of translation initiation sites. It is frequently assumed that initiation of translation in prokaryotes is controlled by a ribosome binding site (RBS), a sequence with the Shine-Dalgarno (SD) consensus situated in the 5 UTR. However, [~]30% of the 5,007 genomes, representing the RefSeq collection of prokaryotic genomes, have either non-SD RBS sequences or no RBS site due to physical absence of the 5 UTR (the case of leaderless transcription). Predictions of the gene 3 ends are much more accurate; still, errors could occur due to the use of incorrect genetic code. Hence, an effective gene finding algorithm would identify true genetic code in a process of the sequence analysis. In this work prediction of gene starts was improved by inferring the GC content dependent generating functions for RBS sequences as well as for promoter sequences involved in leaderless transcription. An additional feature of the algorithm was the ability to identify alternative genetic code defined by a reassignment of the TGA stop codon (the only stop codon reassignment type known in prokaryotes). It was demonstrated that MetaGeneMark-2 made more accurate gene predictions in metagenomic sequences than several existing state-of-the-art tools.
- organism: bacteria (other)
- assay: metagenomics
- research focus: gene regulatory network inference
- approach: statistical model
Abstract summary
Claude Opus 5.5 · xhigh effort · 18 Oct 2026 · from the abstract · CC BY-NC-ND 4.0
Gene caller that models GC-dependent ribosome binding and leaderless-promoter signals and detects TGA stop-codon reassignment, reported as more accurate than existing metagenomic gene finders.
Metagenomic gene callers commonly assume translation starts at a Shine-Dalgarno (SD) ribosome binding site (RBS), yet around 30% of the 5,007 prokaryotic RefSeq genomes surveyed have non-SD sites or no 5' untranslated region at all. Treating both RBS and leaderless-transcription promoters as GC-dependent signals, and inferring whether TGA has been reassigned, targets the two error sources the authors single out: misplaced starts and a wrong genetic code.
Key results
ClaimedMetaGeneMark-2 is reported to predict genes in metagenomic sequences more accurately than several current gene-finding tools.
The abstract asserts a comparative demonstration but names neither the competing tools, the test sequences, nor any accuracy figure. · Abstract
ClaimedStart-codon placement was improved by inferring GC-content-dependent generating functions for RBS sequences and for promoters driving leaderless transcription, rather than assuming a single SD-type signal.
Described as the main modelling change; the abstract reports no before/after accuracy numbers for start prediction. · Abstract
SuggestedRoughly 30% of prokaryotic reference genomes do not fit the standard SD model, motivating the extra signal types; the algorithm also infers whether TGA has been reassigned from a stop codon.
Stated as [~]30% of 5,007 RefSeq prokaryotic genomes having non-SD RBS or no 5' UTR; TGA reassignment is described as the only stop-codon reassignment known in prokaryotes. · Abstract
Methods, data and reproducibility
A protein-coding gene finder for metagenomic contigs in which GC-content-dependent generating functions are inferred for ribosome binding site sequences and for promoter sequences used in leaderless transcription, with an added step that detects an alternative genetic code arising from reassignment of the TGA stop codon. Accuracy was then compared against other gene prediction tools.
- Prokaryotic reference genomes surveyed for RBS type and presence of a 5' UTR · prokaryotes · n = 5,007 genomes
- Metagenomic sequences used for the comparison against other gene finders (not described further)
- Methods & tools
- MetaGeneMark-2, RefSeq
Limitations (2)
- No accuracy numbers, evaluation datasets or names of the compared gene finders are given, so the claimed improvement cannot be judged from the text available. · noted in this summary
- Alternative genetic code detection covers only TGA reassignment, which the abstract describes as the sole reassignment type known in prokaryotes; other deviations would not be caught. · noted in this summary
Read this if you call genes in metagenomic assemblies and need start sites handled for organisms lacking a Shine-Dalgarno motif, or stop codons handled under TGA reassignment.
- Citations
- 49
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Improved Prokaryotic Gene Prediction Yields Insights into Transcription and Translation Mechanisms on Whole Genome Scale
Alexandre Lomsadze, Karl Gemayel, … Mark Yu. Borodovsky·Genome ResearchPeer-reviewed·Jul 1, 2018
- StartLink+: Prediction of Gene Starts in Prokaryotic Genomes by an Algorithm Integrating Independent Sources of Evidence
Karl Gemayel, Alexandre Lomsadze, Mark Yu. Borodovsky·bioRxiv·Oct 25, 2020
- gTranslate: rapid and accurate translation table prediction for prokaryotic genomes
Pierre-Alain Chaumeil, Philip Hugenholtz, Donovan H. Parks·bioRxiv·May 28, 2026
- DeepCDS: Ab initio coding sequence prediction in prokaryotic short nucleotide sequences
Line Sandvad Nielsen, Henrik Nielsen, Ole Winther·bioRxiv·Sep 24, 2026
- A Unique Ribosome Signature Reveals Bacterial Translation Initiation Sites
Adam P Giess, Veronique Jonckheere, … Eivind Valen·BMC BiologyPeer-reviewed·Aug 30, 2017
- Balrog: A universal protein model for prokaryotic gene prediction
Markus J. Sommer, Steven L. Salzberg·PLoS Computational BiologyPeer-reviewed·Feb 26, 2021