Experimental Pecan Run using Gigas genome

I restarted Pecan using 1 of my geoduck .mzML files, and the full digested c. gigas proteome.

It was simple to start, and notice that I was successfully able to use the --backgroundProteome input! The concern is, as always, whether Emu has enough memory to complete all 80 isolation scheme windows for this one file. Here’s a summary:

  1. Created new folder containing the following:
    • Isolation scheme (.csv file)
    • .mzML data file
    • .txt file with path to .mzML data file
    • .txt file with path to background proteome/database
    • Digested c. gigas proteome, with the PRTC standards added
  2. Navigated to the newly created folder containing all Pecan inputs and entered the following:

pecanpie -o ~/Documents/Laura/DNR_geoduck/Pecan7_output/ --backgroundProteome Combined-digested-edited-gigas-QC_Mass400to6000.tabular -n DNR_geoduck_gigasgenome --pecanMemRequest 10 /home/srlab/Documents/Laura/DNR_geoduck/Pecan-inputs7/DNR_Geoduck_mzMLpath.txt /home/srlab/Documents/Laura/DNR_geoduck/Pecan-inputs7/DNR_Geoduck_DatabasePath.txt /home/srlab/Documents/Laura/DNR_geoduck/Pecan-inputs7/DNR_Geoduck_IsolationScheme.csv --fido --jointPercolator --isolationSchemeType BOARDER --overwrite

User input --pecanMemRequest 10.00G is smaller than empirical estimate 136.01G Would you like to use the empirical estimate instead? (Y/N) N

  1. Navigated to the pecanpie generated directory and executed the run_search.sh file:
    cd ../Pecan7_output/ ./run_search.sh Your job-array 630.1-80:1 ("pecan_x104db") has been submitted Your job 631 ("prcltr_x104db") has been submitted Your job 632 ("jPrcltr_x156cc") has been submitted Your job 633 ("pecan2blib.job") has been submitted

  2. Navigated to the three pecan directories to make all .job files executable via `chmod +x [.job file]

  3. Checked to make sure pecan jobs are queued and running: qstat -f

Written on March 17, 2017