Signature reversion and drug analyses

[lizzyjoan]

Summary

These scripts include the initial signature reversion analysis:

• 11_signature_reversion.R the SETBP1-HD consensus signature is used with signatureSearch to query and calculate top drug reversers in brain-relevant cell lines
• 12_drugtarget_degs.R signature reversion drug candidates are analyzed for their targets
• 13_drug_safety.py signature reversion drug candidates are used to query safety data via DailyMed's API
• figure_3.R plotting results from drug analyses and compiling Figure 3

Author Checklist

• This repository has a Lasseigne Lab ruleset configured.
  See instructions here.
• A singularity container has been created with all dependancies and the
  cap_container command has been included in each script using a container.
• All of the necessary scripts, dependencies, and inputs have been included
  in this pull request.
• Only files relevant to this change are included; no temporary, generated,
  or unrelated files are part of this pull request. Use .gitignore to prevent
  irrelevant files from accidentally being included.
• Results have been reproduced and verified with lasseignelab and
  default environments by running the job until two consecutive runs produce
  the same results. Code comments (per script) have been added for
  non-reproducible scripts. Include information about why the script is not
  reproducible.
• All automated code checks have passed. The "All checks have passed"
  message should show in the GitHub pull request status box above the
  "Merge pull request" button.
• No merge conflicts exist. Merge conflicts are indicated by the "This
  branch has conflicts that must be resolved" message in the GitHub pull
  request status box above the "Merge pull request" button.
• Primary and secondary reviewers have been requested to review this
  pull request.

Self/Peer Review Checklist (Coding Guidelines)

• Meaningful variable and function names
• File header comments
• Function comments
• In-line comments summarize logical sections of code by concisely explaining
  why, not what the code is doing. Avoid excessive or redundant commenting.
• Random seed is set
• Reasonable processing time for each job or script is specified
• All outputs have been saved
• Session info included in the scripts
• Cross-reference the working draft of the paper
• Confirm the code performs the intended analysis
• Confirm the code reproduces the reported results

Setup

Make sure you're on cap version 1.0

cd $USER_SCRATCH/setbp1_hd
git checkout main
git pull
git checkout sigsearch

Reproduce results

These scripts must be ran in order and cannot be run at the same time (each one depends on files in the prior script).
Time approximations do not include pulling docker image and converting to a sif file if it's not already available! This may add 10-20 minutes to the run time.

1) Run the signature reversion script (~10-15 minutes)

cap run -s batch src/11_job_sigsearch.sh

2) Run the drug target analysis script (~1-2 minutes)

cap run -s batch src/12_job_drugtarget_degs.sh

3) Run script to pull drug safety data from DailyMed API (~10-15 minutes)

cap run -s batch -e default src/13_job_drug_safety.sh

STOP

Make sure the drug safety script ran fine. You should see an output, but it will not reproduce.

ls -hal data/dailymed

If everything ran fine, then you can remove your copy and symlink my copy:

rm -rf data/dailymed
cap run -s batch -e lasseignelab src/13_job_drug_safety.sh -n
ls -hal data/dailymed

data/dailymed should now be symlinked to the copy in PROJECT_dir in the Lasseigne Lab Cheaha directory.

4) Run plotting script for figure 5 ( < 2 minutes)

cap run -s batch src/figures/job_figure_5.sh

Verification

Run verification script

cap verify verifications/11_sigsearch.sh

git diff verifications/11_sigsearch.out

If nothing is returned, the outputs match!

Note: The ./results/signature_reversion/drug_safety_data.csv is not reproducible because it's using the DailyMed API which updates daily. The script has a comment noting this, and it is excluded from the md5 verification. The specific version used will be added to Zenodo

[lizzyjoan]

@tsoelter I updated the cap verification and also had to pull updates from main after the consensus signature was merged, so it dismissed your review

[tsoelter]

The updated verifications and all changes work perfectly! 🎉

[lizzyjoan]

Putting review on hold--need to add "pediatric" drug label excerpts for dailymed query and update runs with new prioritized SETBP1 targets

[sashataluri]

The following command ls -hal data/dailymed produces this error for me: ls: cannot access data/dailymed: No such file or directory

Not sure why I am getting it. Maybe an issue on my end. Everything works fine otherwise, and the code looks good! I verified the MD5 sums, which were clear. I can approve it when you get a chance to look at this.

[tsoelter]

Everything still looks good, works, and verifies on my end! 🎉

[lizzyjoan]

The following command ls -hal data/dailymed produces this error for me: ls: cannot access data/dailymed: No such file or directory

Not sure why I am getting it. Maybe an issue on my end. Everything works fine otherwise, and the code looks good! I verified the MD5 sums, which were clear. I can approve it when you get a chance to look at this.

Hmm, can you check your log files for the 13_job_drug_safety run? (i.e.,

cd $USER_SCRATCH/setbp1_hd
cat logs/13_job_drug_safety*

and let me know what it shows?