README.md

metadata

pretty_name: GAMBA ATG Contexts
license: mit
tags:
  - biology
  - genomics
  - dna
  - genome-language-model
  - benchmark
  - translation-initiation
  - hg38
  - parquet
configs:
  - config_name: causal
    data_files:
      - split: test
        path: atg-gamba-causal.parquet
  - config_name: bidi
    data_files:
      - split: test
        path: atg-gamba-bidi.parquet

GAMBA ATG contexts

This dataset packages GAMBA's five source ATG contexts. GAMBA selects 2,000 complete MANE Select transcript examples approximately evenly across chr1–chr22; each example contributes five rows:

1. true translation start;
2. nearby noncoding ATG, 2–5 kb away;
3. far noncoding ATG, at least 100 kb away;
4. same-transcript in-frame methionine;
5. same-transcript out-of-frame ATG motif.

The five rows share a pair_id, yielding 10,000 rows total.

The paper leaderboard is four-class: merge noncoding_near and noncoding_far into one noncoding label, then use cosine leave-one-out 1-nearest-neighbor balanced accuracy. GAMBA's extraction script also writes a separate five-way Euclidean diagnostic; that is not the reported leaderboard.

Choosing a context

The GAMBA paper uses context geometry matched to the model. Choose causal for Evo2 or another left-to-right model: the ATG is at the end of the strand-oriented window, so all preceding bases are usable context. Choose bidi for GAMBA encoders, the distilled student, GPN-Star, PhyloGPN, or another masked/bidirectional model: the ATG is centered so the model sees both flanks. Using a causal file for a bidirectional model removes its right-side context and is not the paper protocol.

Loading

from datasets import load_dataset

dataset = load_dataset("Taykhoom/atg-gamba", "bidi", split="test")

Columns

Column	Description
split, label, label_id	Always test and one of the five ATG classes.
sequence	Strand-oriented causal or bidirectional context of at most 2,048 bp.
context_policy	causal for causal; symmetric for bidi.
pair_id	Shared identifier for all five sites from one transcript example.
category, scope	Constant ATG category and ROI pooling scope.
chrom, start, end, strand	Zero-based, half-open hg38 three-base site coordinates.
context_start, context_end	Forward-genome coordinates of sequence.
roi_start, roi_end, pool_start_in_window, pool_end_in_window	ATG offsets within the context.
name, transcript_id, gene_id	Site and source transcript identifiers.
delta_bp	Signed distance from the true start; zero for label 1.
phylop_mean, phylop_std	Mean and population standard deviation over the three-base ATG.
phylop_frac_pos, phylop_frac_neg	Fractions of ATG bases with positive or negative phyloP scores.
phylop_mean_pos, phylop_mean_neg	Means over positive or negative ATG scores, or zero when absent.
phylop_context_mean, phylop_context_std	Mean and population standard deviation over GAMBA's symmetric 2,048 bp phyloP context.
phylop_context_frac_pos, phylop_context_frac_neg	Fractions of context bases with positive or negative scores.
phylop_context_mean_pos, phylop_context_mean_neg	Means over positive or negative context scores, or zero when absent.

Data source and processing

The chromosome-level source TSVs, combined TSV, and sampled 2,000 examples produced by Gamba-Processing match Microsoft GAMBA commit e83984e byte-for-byte.

GAMBA stores minus-strand ATG sites as [position, position + 3) before reverse complementation. This release preserves that convention exactly.

PhyloP values come from the Zoonomia 241-mammalian track used by GAMBA. Uncovered positions are zero and scores are rounded to two decimal places before the float32 summaries are computed.

Citation

Please cite the GAMBA paper:

Consens, M. E. et al. Predicting evolutionary rate as a pretraining task improves genome language model representations. bioRxiv (2026). doi:10.64898/2026.02.02.703275. A conference version is available on OpenReview.

@article{consens2026predicting,
  title = {Predicting evolutionary rate as a pretraining task improves genome language model representations},
  author = {Consens, Micaela Elisa and Yang, Kevin K. and Hall, Jimmy and Conard, Ashley Mae and Wang, Bo and Crawford, Lorin and Moses, Alan and Lu, Alex X.},
  journal = {bioRxiv},
  year = {2026},
  doi = {10.64898/2026.02.02.703275},
  url = {https://doi.org/10.64898/2026.02.02.703275}
}

License

GAMBA and its released benchmark code/data are distributed under the MIT license. Upstream annotation resources retain their original terms.