Loading…
Reference-free enzyme class annotation of single unassembled metagenomic reads, assigning top-level EC classes without assembly or homology search.
No providers recorded yet. Browse all providers
Metagenomic functional annotation conventionally assembles reads into genes and then matches those genes to a reference database. Both steps discard information. Assembly fails in low-coverage or highly diverse communities, and reference matching cannot describe anything without an annotated relative — a severe constraint in extreme environments, where alignment pipelines label as little as 0.2% of reads.
REBEAN (Read Embedding-Based Enzyme ANnotator), from Yana Bromberg's group at Emory University, skips both steps. Given one raw sequencing read, it predicts which of the seven top-level Enzyme Commission classes the gene that read came from encodes, or that the gene is not an enzyme at all. The framing is deliberate: recognize the function, do not identify the gene. Nothing is aligned and nothing is assembled.
It is a fine-tune of REMME, the read-level DNA language model released with it, which supplies the sequence representation that makes reference-free prediction possible. The work was posted to bioRxiv in December 2024 and published in Nucleic Acids Research in September 2025.
The model adds a three-layer dense classifier over the pretrained encoder. Fine-tuning used 19 million non-redundant reads drawn from 19,316 metagenomic samples across nine environments, with training labels generated by mi-faser, a high-precision alignment method. It attained 80.6% test accuracy, a macro AUC of 0.969, recall of 80.1%, and precision of 83.6% across eight classes. Training the same architecture from random initialization plateaued at 30% accuracy, which is the clearest evidence that the pretraining does the heavy lifting rather than the classifier head.
On an independent set of 525,775 synthetic reads from 4,295 experimentally annotated prokaryotic enzymes, accuracy was 88.1%, with precision of 71.5% and recall of 33.4% at a 0.5 threshold — the gap between those two recall figures being the honest measure of how much the mi-faser-derived training labels flatter the held-out split.
The intended use is functional profiling of metagenomes that alignment pipelines handle poorly. On 8 million reads from hydrothermal vents, hypersaline lakes, and salt crystallizer ponds, the model annotated 6.1% of reads as enzymatic at a 90% precision threshold — three- to six-fold more than alignment-based tools, and thirty-fold more than HUMAnN3 — while agreeing closely with those tools where they did produce calls (Cohen's kappa 0.945 against mi-faser, 0.880 against DIAMOND). Applied to 3,820 marine metagenome-assembled genomes, it flagged 4,901 proteins as putative oxidoreductases at 90% precision, enriched for oxidoreductase-associated Pfams at an odds ratio of 3.66, including candidates with no close structural match in the PDB.
By labeling function rather than identity, this makes reference-free annotation possible for the metagenomic data that assembly and homology search leave untouched — which, in extreme environments, is nearly all of it. The model is small enough to run on modest hardware, and the release includes weights, scripts, Figshare archives, and a hosted service.
Limitations are acknowledged and material. Training labels come from mi-faser rather than experiment, so reported accuracy partly measures agreement with a precise but low-recall aligner; recall on non-enzymatic reads is only 50.6%; and prediction stops at the seven top-level EC classes rather than naming a specific reaction, so a call is a starting point for investigation rather than an annotation in the usual sense.
Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.