Dissection of core promoter syntax through single nucleotide resolution modeling of transcription initiation (CLIPNET models)

Model weights for CLIPNET. Chromosomes in fold 0 were held out for all models. The 9 remaining folds were used to train models such that data fold i was withheld from training for model i (for assignments, see data_fold_assignments.csv). The models assume a 2-hot encoding of DNA (A=[2, 0, 0, 0], C=[0, 2, 0, 0], ..., Y=[0, 1, 0, 1], ...).

Processed data used for model training and to reproduce figures in our CLIPNET paper (preprint here) are archived at DOI 10.5281/zenodo.10597357

Model weights are also available for download via HuggingFace.