CaPTk Feature Extraction Error on Cluster

[sherjeel275]

Describe the bug I am running the FeatureExtraction module of CaPTk on CBICA Cluster, but the output I'm getting is significantly different from the output I get when I run the FeatureExtraction locally on my pc. The output from the cluster has numerous variables that are different. I'm not sure why the variables being generated on the cluster are different.

To Reproduce Steps to reproduce the behavior:

1. Run FeatureExtraction module on local pc
2. Run FeatureExtraction module on cluster
3. Compare the two outputs

Cluster Command: FeatureExtraction -b batch.csv -p params.csv -o output.csv

Local Command: C:\CaPTk_Full\1.8.1\bin\FeatureExtraction -b batch.csv -p params.csv -o output.csv

Expected behavior The 2 outputs should be the same and should have the exact same features inside the csv files that get generated.

CaPTk Version I'm using version 1.8.1

Desktop (please complete the following information):

• OS: Windows

[sarthakpati]

Hey,

What is the version of CaPTk you are running on the cluster? Also, can you post the parameter file you are using?

[sherjeel275]

I think I'm using v1.8.1 on the cluster. I did module load captk/1.8.1 before running FeatureExtraction on the cluster.

[sarthakpati]

Okay. Could you add the following lines before you call FeatureExtraction as a sanity check?

which FeatureExtraction 
FeatureExtraction -v

I have seen your email with the parameter file, thanks.

[sarthakpati]

Hi,,

Based on the data you had sent, I found a single source of difference, which I consider negligible:

outputs.zip

Added the outputs for reference. I extracted the features using v1.8.1 from my local Windows and from the cluster.

Closing the ticket, please re-open if you have further questions.

Cheers, Sarthak