A Reusable Machine-Constructed Feature Representation for Protein Sequences

Multi Purpose Protein Prediction Vector (MP3vec) representation is constructed from a protein’s sequence and its PSSM profile using all the available high resolution protein structural data with the aim to aid in 'transfer learning' to overcome the 'small data' problem in developing predictive models for biomacromolecular interactions. Please refer to the Readme file and the binary can be downloaded as a zip file.