search

DeepRank / iScore

iScore: an MPI supported software for ranking protein-protein docking models based on a random walk graph kernel and support vector machines
Apache License 2.0
26 stars 11 forks source link

a bug with "mpiexec -n ${NPROC} iScore.predict" #21

Closed LilySnow closed 4 years ago

LilySnow commented 4 years ago

When I run "mpiexec -n ${NPROC} iScore.predict" on capri target 161, I ran into a bug.

  1. I run this command for the first time, I will get the following error message:

Reusing kernels in ./kernel/ Traceback (most recent call last): File "/home/lixue/tools/iScore/bin/iScore.predict", line 57, in iscore_svm(load_model=None,package_name=args.archive,testID=args.ground_truth) File "/home/lixue/tools/iScore/iScore/graphrank/rank.py", line 402, in iscore_svm testdata = DataSet(trainID,Kfile,maxlen,testID=testID) File "/home/lixue/tools/iScore/iScore/graphrank/rank.py", line 29, in init self.test_name, self.test_class = self._get_ids(testID) File "/home/lixue/tools/iScore/iScore/graphrank/rank.py", line 80, in _get_ids nc = len(idlist[0]) IndexError: list index out of range

The folder of 'graph' will be calculated successfully but 'kernel' folder will not.

  1. I then deleted the folder of 'kernel', and rerun the same command. This time the kernel folder is calculated correctly. But I got the following error message:

Traceback (most recent call last): File "/home/lixue/tools/iScore/bin/iScore.predict", line 63, in iscore() File "/home/lixue/tools/iScore/iScore/function/score.py", line 17, in init self.read_graphrank() File "/home/lixue/tools/iScore/iScore/function/score.py", line 56, in read_graphran data.append(self.features[m]['grank']) KeyError: 'grank'

  1. Then I ran the same command the 3rd time, iScorePredict.dat is finally generated.

The data is on alcazar: /home/lixue/CAPRI/Capri48/t161/iscore/before_cleaning/results_li

CunliangGeng commented 4 years ago

Without MPI

iScore.predict --archive ../train/training_set.tar.gz

With MPI

mpiexec -n ${NPROC} iScore.predict.mpi --archive ../train/training_set.tar.gz

NicoRenaud commented 4 years ago

To avoid this confusion it might be a good idea to keep only the MPI execs.

CunliangGeng commented 4 years ago

I have updated the README to clearly show how to use them.