Module 5: Additional Notes

Computing clusters, job submissions, and ModelArray

To run on the cluster, we recommend you use our tutorial on cluster-based submissions located at ( https://github.com/PennLINC/hbcd_pipeline_replication_guide/tree/gh-pages/processing_code/mass_univariate ).

How to output brain-based data for visualizations

To output the data for brain-based visualization we must first output the results as an h5 file. Change the /path/to/a/new/h5/file.h5 to output the results to your preferred filename.

writeResults("/path/to/a/new/h5/file.h5", df.output = res_wrap, analysis_name = "mapmri_rtop_anova_and_means")

Next, we will need to convert the h5 file into a NIFTI (.nii.gz) file. We can modify the code chunk for voxelstats_write to output the NIFTI. We must provide the correct cohort-file for non-imaging data, analysis-name saved above, the correct input-hdf5 imaging input file, and an example-nifti template for creating the NIFTI output. output-dir can be modified towards where the output NIFTI will be located.

$ voxelstats_write \
  --cohort-file /nbdc-sandbox/derivatives/qsirecon-modelarray/ABCC_mapmri_rtop_phenotypes.csv \
  --analysis-name mapmri_rtop_anova_and_means \
  --input-hdf5 /nbdc-sandbox/derivatives/qsirecon-modelarray/qsirecon-mapmri_rtop.h5 \
  --output-dir /where/you/want/to/put/your/niftis \
--example-nifti /nbdc-sandbox/derivatives/qsirecon-modelarray/qsirecon-mapmri_rtop.nii.gz # Need to find the real path