neuroPID

NeuroPID is a prediction tool for Neuropeptide Precursor (NPP) and Neuromodulator Proteins. NeuroPID provides a list of candidate NPPs and neuromodulators at a genomic scale from unexplored proteomes using protein FASTA sequences as input.

Authors

• Original package: Dan Ofer at ddofer"at"gmail.com.
• Refactor and modernisation: Fabio Zanini (fabio dot zanini at unsw dot edu dot au).

LICENSE

MIT.

Installation

Packaging is WIP. For now, you can install (e.g. in a venv) the following dependencies:

numpy
biopython
tmdq
scikit-learn

The package seems to be liking numba but it's unclear at this point whether it actually uses it for anything useful.

Repo file tree

• README.md: This file.

• LICENSE: License file (MIT).

• neuropid: Source files

  • get_fastasets.py:

    • Downloads Neuropeptides from uniprotKB to a single multi-FASTA file, and a length-binned distribution of Negatives (non Neuropeptides) into multiple multi-FASTAs.

  • local_SLEEK_FeatureGen+_new:

    • Extracts feature data from fasta file(s) in same directory as it, outputs results to two tsv files (one for actual neuropeptides, one for negative control sequences).

  • train_classifier.py:

    • Train and tests various ML classifiers. Requires prior generation (via FeatureGen+) of Feature data .txt files for training, and from the target/test (multi_fasta)!
    • Reads/Imports the +- Training sets' feature data from a predefined location (must be entered in the script, or you can change the code parameters to accept user inputted dir/location instead).
    • After Training data is imported, ML is automatically trained on it, then performs prediction on a given target/test file (containing feature data) location (for prediction).

  • getTopPredictedOrganismResults.py (not refactored yet):

    • Similar to Testing_organismsML, outputs the " best" results that have a probability/quorum past a user defined threshhold, and outputs the names of the samples that met this threshhold into a CSV file.

  • Model_Statisticalparameters_Calc.py (not refactored yet):

    • Used to test performance of various paramters and schemes used for machine learning and the data sets.

• data: Folder for data files used for testing etc.

• results (not refactored yet): Looks like an old web thing, doubt it still works.