Data Sets and Results for "Improved data sets and evaluation methods for the automatic prediction of DNA-binding proteins"

Data sets and results for "Improved data sets and evaluation methods for the automatic prediction of DNA-binding proteins"

The file "dna_binding_protein_sequences.zip" has the training and testing sets from the paper:

RLL - "random_<train/test>_full_1000.csv"
RSL - "random_<train/test>_40.csv"
RS&LL - "random_<train/test>_40_1000.csv"
RLL where included positive examples have verified DNA binding activity - "random_<train/test>_hq_1000.csv"
The 10 RS&LL data sets - "random_<train/test>_40_1000.csv" + "random_<train/test>_40_1000_cv_<0-8>.csv"

The results files are named similarly. See "see_results.ipynb" in the codebase that supplement these data sets
The species data sets are derived from "uniprot_data_bac.tab" and "uniprot_data_not_bac.tab."  See code. 

The ESM embeddings used by the XGBoost model are in "dna_binding_protein_esm.zip"