Data and Weights for Reverse Homology
Bibliographic record
Abstract
Training data, weights, and classification datasets for "Discovering molecular features of intrinsically disordered regions by using evolution for contrastive learning". Training data: scer_idr_homologues and human_idr_homologues contain a zip file of the fasta files of IDR homologues used to train the yeast and human model, respectively. Note that these fasta files are aligned, but we strip away the alignment symbol "-" before input into our model. disprot_idr_homologues contain a zip file of IDR homologues corresponding to the DisProt database. human_protein_alignments contains fasta files for the full proteins containing these IDRs Weights: scer_idr_model and human_idr_model contain a zip file of the weights for the yeast and human model respectively, which can be loaded into the model files at github.com/alexxijielu/reverse_homology. Likewise, disprot_idr_model contains a zip file of our model trained on DisProt IDRs exclusively. Logo Websites: scer_idr_logo_website, human_idr_logo_website, and disprot_idr_logo_website contain a zip file of an HTML file for the yeast, human, and DisProt model respectively, showing sequence logos of the features learned by each model and their enrichments. Features: human_idr_features contains the raw feature activations for all human IDRs in our human model. (We didn't include this file in the supplementary for the paper due to size.) Classification datasets: IDR_classification_datasets contains datasets used in our benchmarks. These datasets are encoded as binary csv matrixes. cdc28_classification contains IDRs labeled as Cdc28 phosphorylation sites, mitochondrial_targeting_classification contains IDRs labeled as mitochondrial targeting signals, evosig_cluster_classification contains IDRs labeled by clusters assigned in previous computational work by Zarin et al. eLife 2019, and go_SLIM_classification contains proteins labeled by GO Slim annotations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.007 | 0.017 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.053 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; both teacher heads agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".