MaxQuant is a quantitative proteomics software package designed for analyzing large mass-spectrometric data sets. This demo gives instructions for running the MaxQuant proteomics
software on Fred Hutch’s
Gizmo compute cluster.
Note: The process described here replaces the previous cloud- and Windows-based pipeline, which is now deprecated. That pipeline will no longer be maintained and will eventually be disabled.
Our on-premises compute cluster (Gizmo) can do the compute-intensive work of running MaxQuant. You still need to run MaxQuant on your local Windows machine to configure your job, but instead of running it locally (or in the cloud), you will run it on the compute cluster.
If you have used the previous (cloud-based) pipeline, you already know how to run MaxQuant on a local Windows machine to configure your job.
The process is described here.
If all goes well in this step, you will end up with a directory on the
X: drive (accessible from both Windows and Linux) which contains your
mqpar.xml file, your
.raw data files, a
.fasta file and optional
Edit XML file
You now need to edit the XML file (
mqpar.xml) in a text editor.
Visual Studio Code is a good choice if you don’t have a favorite.
You need to make the following changes (note that in future, there may be an automated submission script that takes care of this for you, but for now it’s a manual process).
Convert Windows paths to Linux paths
<filePaths> in the
mqpar.xml file to find the paths to your data files. Also search for
<fastaFilePath> to find the path to your
Note that the
X: drive on Windows is equivalent to the
/fh/fast filesystem on Linux.
You need to convert all Windows paths to Linux, which means:
- Replace the Windows drive letter (
X:, etc.) with
- Replace all backslashes (
\) with forward slashes (
Example: You may have paths like this in your
<filePaths> <string>c:\mq-job\JK050120_050620_Ub_BioRep1.raw</string> <string>c:\mq-job\JK050120_050620_Ub_BioRep2.raw</string> <string>c:\mq-job\JK050120_050620_Ub_BioRep3.raw</string> </filePaths> ... <fastaFilePath>c:\mq-job\uniprot-proteome_UP000005640.fasta</fastaFilePath>
If your PI is Jane Doe, you will want to change these paths to something like the following (replace
doe_j with your actual PI’s last name and first initial):
<filePaths> <string>/fh/fast/doe_j/proteomics/maxquant-jobs/job20200610/JK050120_050620_Ub_BioRep1.raw</string> <string>/fh/fast/doe_j/proteomics/maxquant-jobs/job20200610/JK050120_050620_Ub_BioRep2.raw</string> <string>/fh/fast/doe_j/proteomics/maxquant-jobs/job20200610/JK050120_050620_Ub_BioRep3.raw</string> </filePaths> ... <fastaFilePath>/fh/fast/doe_j/proteomics/maxquant-jobs/job20200610/uniprot-proteome_UP000005640.fasta</fastaFilePath>
Again, you need to make sure that the directory
/fh/fast/doe_j/proteomics/maxquant-jobs/job20200610/, equivalent to
X:\doe_j\proteomics\maxquant-jobs\job20200610\ on Windows) actually exists and contains your
mqpar.xml file, your data files, your
.fasta file, and any other optional
Change .NET setting
In MaxQuant 18.104.22.168 and later, you need to make sure that the setting to use .NET Core is turned off (which it may not be by default), otherwise you may encounter an error in your run.
Search for the string
useDotNetCore in your
If the string is present, change
If the string is not present, add the above line somewhere in the
MaxQuantParams section of your XML file.
A good place to put this line is in between
Change the number of threads to match the number of data files
The Windows-based pipeline took care of this step automatically, but it’s a quick manual edit. Search for
<numThreads> and change the its value to the number of
data files you have.
For example you may see this:
If you have 24 data files in your job, change the number accordingly:
Remember the number you changed it to; you will need it later.
NOTE If you have more than 36 data files, set this number to 36.
Submit job to cluster
The rest of these steps need to be done on a Linux HPC system. Using
NoMachine, connect to the system
rhino01. For more information about these connection methods, read Computing Resource Access Methods.
Change to job directory
Let’s go to the directory where our xml, fasta, and data files are.
In the earlier example, that directory was
Substitute your PI’s last name and first initial for
doe_j. The rest of the path after the PI’s name is arbitrary and can be whatever you choose. But it must match the paths in the XML file that you edited, and all your files (
.fasta and data files) must exist there.
So to change to the directory, use
List MaxQuant modules
Let’s check and see which versions of MaxQuant are available with the command
ml avail maxquant:
$ ml avail maxquant ------------------------------- /app/modules/all ------------------------------- MaxQuant/22.214.171.124-foss-2018b Use "module spider" to find all possible modules and extensions. Use "module keyword key1 key2 ..." to search for all possible modules matching any of the "keys".
This shows us that there is one MaxQuant module available and it is called
NOTE: If you need a newer version of maxquant made available on the cluster, please contact
Create submit script
Create a script with the following contents using your favorite text editor and make sure it ends up in the same directory as your
mqpar.xml file and data files. Call the script
#!/usr/bin/env bash # Send me email when the job is done. #SBATCH --mail-type=END # Number of tasks; we just want one: #SBATCH -n 1 # Number of cores per task, CHANGE THIS: #SBATCH -c 24 # Amount of memory to use. CHANGE THIS to what you need: #SBATCH --mem 250G # Number of days to let the job run. CHANGE THIS. # In this case, 3 is the number to change: #SBATCH -t 3-0 source /app/lmod/lmod/init/bash module use /app/easybuild/modules/all ulimit -n 655360 # Here's the name of the MaxQuant module to load. # This may change over time as new versions become available. # On rhino01, you can run the command # `ml avail maxquant` to see the available versions. ml MaxQuant/126.96.36.199-foss-2018b maxquantcmd mqpar.xml
Be sure and change all values where it says
CHANGE THIS, unless the values set are ok with you.
For CPU cores, set this to the number of data files you have (unless you have more than 36 data files, in which case, set this to 36).
Now, on rhino01, after changing to the directory where your XML and data files live, type the following command:
chmod +x submit.sh
That makes your script executable.
Then submit your script to the cluster:
sbatch will print out a job number, like this:
Submitted batch job 48886585
Hold on to that job number, it will be useful for checking on the status of your job.
At this point, you can safely disconnect from
rhino01. Your job is running (it will take hours or days to complete) and ending your
NoMachine session will not stop it from completing. But let’s stay connected for a moment to learn a couple other things.
If you run
ls now, you will notice that there is a log file whose name contains the job number (jn this case it is
slurm-48886585.out). One trivial way to monitor job progress is to run
tail -f slurm-48886585.out
This will watch that file for changes and show you new lines as they are appended to the end. Type
control-c to exit from the
tail command (it won’t exit automatically when your job is done).
You can also use the
squeue command to check on the status of your job:
$ squeue --job 48886585 JOBID PARTITION NAME USER ST TIME NODES NODELIST(REASON) 48886585 campus-ne submit.s dtenenba R 0:07 1 gizmok30
ST stands for “status” and a value of
R for that column means the job is running.
If this command does not return anything, it means your job has completed (either successfully or not). Check the log file as described above for further information.
maxquant will run for hours or days and when it is done you will receive a blank email from
slurm user with a subject line like this:
Slurm Job_id=48886649 Name=submit.sh Ended, Run time 00:00:01, COMPLETED, ExitCode 0
maxquant is finished, your output files will be in this same directory, and you can access them from Windows or Linux.
Check the Gizmo documentation on this wiki for further information on running batch jobs on the Gizmo cluster.
If you have any questions, please contact
Updated: November 23, 2020Edit this Page via GitHub Comment by Filing an Issue Have Questions? Ask them here.