/ATP-Interaction

Predicting ATP interacting residues by training a model using SVM scikit learn

Primary LanguagePython

AIM: Prediction of ATP interacting residues

HOW TO RUN

`python3 script.py`

STEPS

  • TRAINING

    • Choose a value of n, i.e length of patterns. eg, n=7/n=9,n=17 etc.

    • Add ((n-1)/2)'XXX..' in the starting and end of the protien sequence such that we have (XX..protien_seq..XX).

    • Make patterns or cut windows of length n for each protien sequence.

    • Apply Binary Profiling on each pattern i.e we'll have matrix of size (21 * n) for one pattern Binary profile essentially means making a 2-Dimentional array, which stores 1 for each match with the list of Amino Acids and 'X'. Else, stores 0 for non-match.

    • We know that a small aplhabet indicates an ATP interacting residue, so mark it as -1 (non interacting) or +1 (interacting)

    • Used OneVsRestClassifier in SVM.

    • Used (.fit(X,Y)) to train the dataset, where X is the binary profiled matrix and Y is the list containing answers i.e if each residue is ATP interacting or not.

  • TESTING

    • Predict using predict(X) function by passing the residue protien as an argument.

    • For Testing purposes, we join all amino Acids into a protien and then perform pattern formation (as explained above in training dataset) followed by binary profiling.

OUTPUT

  • Output.txt with ID,Label columns, where label is value

    • +1 = ATP INTERACTING RESIDUE
    • -1 = NON ATP INTERACTING RESIDUE

  • Kaggle score = 0.61957

Output file

Identification of ATP binding residues of a protein from its primary sequence

Chauhan, J.S., Mishra, N.K. & Raghava, G.P. Identification of ATP binding residues of a protein from its primary sequence. BMC Bioinformatics 10, 434 (2009). https://doi.org/10.1186/1471-2105-10-434