PMut

PMut2017

Predictor information

Name	PMut2017
Description	PMut2017 predictor is trained using SwissVar (Humsavar) as a training set. It is built using the PyMut library, with a Random Forest classifier based on 12 features.
Created	Dec. 30, 2016, 2:49 p.m.
Number of proteins	12,141
Number of variants	65,281: 38,078 neutral / 27,203 disease

Log Completed

[2016-12-31 08:48:47] Checking computation status
[2016-12-31 08:50:55] Computing missing alignments
[2016-12-31 08:50:55] Computing features for the variants
[2016-12-31 08:52:02] Features of 65281 variants already computed.
[2016-12-31 08:52:02] Drawing feature distribution plots
[2016-12-31 08:55:09] Training classifier
[2016-12-31 08:57:24] Evaluating classifier

Step 1. Multiple Sequence Alignments

100% Done 12,141 of 12,141 BLAST UniRef100
100% Done 12,141 of 12,141 BLAST UniRef90
100% Done 12,141 of 12,141 Kalign UniRef100
100% Done 12,141 of 12,141 Kalign UniRef90

Step 2. Features

100% Done 65,281 of 65,281 Features computed
100% Done 12 of 12 Feature distribution charts

Step 3. Classifier training

Done Train classifier

Step 4. Classifier evaluation

Done k-fold cross-validation
Done k-fold cross-validation with 50% sequence identity exclusion