Downloads · 30 days
12
27% of all-time downloads
bio-ontology-research-group/otu-taxa
otu-taxa is a machine learning model from bio-ontology-research-group. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for pytorch. The card lists the license as other.
This repository contains the files needed to reuse the pretrained OTU-Taxa model from the paper "OTU-Taxa: a taxonomy-aware foundation model for microbiome samples".
Downloads · 30 days
12
27% of all-time downloads
All-time downloads
44
Public
Repo size
335 MB
Likes
0
Public
Click a slice to open those files.
.pt271 MB · 80%
From the Hugging Face model README
This repository contains the files needed to reuse the pretrained OTU-Taxa model from the paper "OTU-Taxa: a taxonomy-aware foundation model for microbiome samples".
Hugging Face repository:
bio-ontology-research-group/otu-taxa
OTU-Taxa represents each microbiome sample as an abundance-ordered sequence of paired OTU and taxonomy tokens. The released checkpoint is an autoregressive Transformer trained to predict the next OTU-taxonomy pair. The paper model uses SINTAX-derived taxonomy labels and a hierarchy-aware taxonomy path loss with rank-specific unknown taxonomy nodes.
Raw abundance values are not direct model inputs. Abundance is encoded through the within-sample ordering of OTU-taxonomy pairs.
config.json
model_config.json
special_tokens.json
otu_vocab.json
taxonomy_vocab.json
taxonomy_tree.json
checkpoints/best.pt
checkpoints/meta.json
checkpoints/split.json
checkpoints/otu2tax.npy
checkpoints/metrics.jsonl
tree_artifacts/taxonomy_vocab_with_unk.json
tree_artifacts/descendant_matrix_with_unk.npy
tree_artifacts/rank_idx.npy
examples/minimal_inference.py
CHECKSUMS.sha256
checkpoints/best.pt is the main pretrained model checkpoint. Keep it paired
with the vocabularies, hierarchy artifacts, and checkpoints/meta.json.
The checkpoint SHA-256 is
689509acc669ff6833e07498cb3e13cf8d6172ac47b1ad881b8990d370ec872b.
Use this model for OTU-Taxa inference, representation extraction, sample classification fine-tuning, generative community readouts, and context-based taxonomy prediction workflows described in the paper.
The source code is maintained in:
https://github.com/bio-ontology-research-group/otu-taxa-autoregressive
This checkpoint depends on the released OTU and taxonomy vocabularies. New data must be processed into the same OTU-token and taxonomy-token conventions or handled with compatible preprocessing code.
The training corpus is derived from the filtered Microbe Atlas Project snapshot used in the paper. Redistribution of raw source data is not included here.
The released OTU-Taxa checkpoint SHA-256 is
689509acc669ff6833e07498cb3e13cf8d6172ac47b1ad881b8990d370ec872b.
Pope-MAP and Waypoint-MAP are comparator models and are not included in this
model repository.