DeepSEA model files

This CNN is based on the DeepSEA model from Zhou and Troyanskaya (2015). The model has been converted to a pytorch model on a modified version of https://github.com/clcarwin/convert_torch_to_pytorch Use this model only for predictions of sequences, not for variant effect prediction. The model automatically generates reverse-complement and averages over forward and reverse-complement to results from the website. To predict variant effects use the DeepSEA/variantEffects model. It categorically predicts 919 cell type-specific epigenetic features from DNA sequence. The model is trained on publicly available ENCODE and Roadmap Epigenomics data and on DNA sequences of size 1000bp. The input of the tensor has to be (N, 4, 1, 1000) for N samples, 1000bp window size and 4 nucleotides. Per sample, 919 probabilities of a specific epigentic feature will be predicted.