ngphubinh

The software tool "ngphubinh" is a GitHub repository that contains the source code and datasets for a novel approach to predict the function of adaptor proteins using recurrent neural networks (RNNs) and position-specific scoring matrix (PSSM) profiles. The main features and findings of this tool are:

It addresses the problem of losing sequential information in PSSMs when using traditional neural networks for protein function prediction.

The tool incorporates RNNs and PSSM profiles to preserve the sequential information and improve the prediction performance.

4. The area under the receiver operating characteristic curve (AUC) metric for predicting adaptor proteins is 0.893 in cross-validation and 0.853 in independent datasets.

Topic

Machine learning;Transcription factors and regulatory sites;Membrane and lipoproteins;Structure prediction

Detail

  • Operation: Post-translation modification site prediction;Pathway or network prediction;Fold recognition;Protein function prediction

  • Software interface: Command-line interface

  • Language: -

  • License: Not stated

  • Cost: Free of charge

  • Version name: -

  • Credit: No funders are listed in the provided text.

  • Input: -

  • Output: -

  • Contact: Susanto Rahardja susantorahardja@ieee.org

  • Collection: -

  • Maturity: -

Publications

  • Classification of adaptor proteins using recurrent neural networks and PSSM profiles.
  • Khanh Le NQ, et al. Classification of adaptor proteins using recurrent neural networks and PSSM profiles. Classification of adaptor proteins using recurrent neural networks and PSSM profiles. 2019; 20:966. doi: 10.1186/s12864-019-6335-4
  • https://doi.org/10.1186/S12864-019-6335-4
  • PMID: 31874633
  • PMC: PMC6929330

Download and documentation

< Back to DB search