Supporting data for "Complete end-to-end learning from protein feature representation to protein interactome inference"

This dataset was created to support the development and evaluation of FREEPII, a deep learning-based framework for predicting protein-protein interactions (PPIs) from co-fractionation mass spectrometry (CF-MS) data. The motivation behind this work is to improve the robustness and generalizability of CF-MS-based PPI predictions by integrating multimodal biological information and leveraging recent advances in deep representation learning. <br>The dataset includes processed CF-MS profiles, protein sequence-based features (including frequency-based encodings and pretrained embeddings), ground truth labels derived from known protein complexes, and the output predictions generated by our model. Also included are the scripts and intermediate files used to generate Figures 25 and Supplementary Figures S1 and S3S10 in the associated manuscript.