MaxQuant version: 2.7.5.0
Module: Tims-MaxDIA
Organism: Drosophila
Library: predicted
Dataset / acquisition
Samples: 8
Fractions per sample: 3
Total raw files: 24
Gradient: 2 hours per fraction
Window design: optimized variable windows generated with py_diAID
MaxQuant settings
Threads: 24
Deafult settings, specifically:
Enzyme: Trypsin
Max missed cleavages (DIA): 1
Default modifications and peptide length settings
No other changes from defaults (happy to share the mqpar.xml if helpful)
Observed runtime
Total processing time was ~18 days.
From runningTimes.txt, the dominant steps were:
DIA library search3: 20,957 min (~14.5 days)
Deisotoping: 1,865 min (~31 hours)
Create slice data: 1,151 min (~19 hours)
Hardware / environment
OS: Windows Server 2019
CPU: 48 cores
RAM: 512 GB
Storage (main run): 22 TB SAS HDD (7.2K RPM, 12 Gbps interface)
Because the runtime was so long, I suspected disk I/O. I repeated a subset run (2 samples) on a boot-optimized M.2 SSD (~480 GB) and compared it to HDD. Both took ~7 days for the same subset, so SSD vs HDD did not materially change the runtime.
Is it expected that “DIA library search3” dominates runtime to this extent for predicted libraries / TIMS DIA?
Given SSD did not improve runtime, what are the likely bottlenecks (CPU, RAM)?
Any recommendations on what to upgrade to reduce runtime if we want to run Tims-MaxDIA routinely?
MaxQuant version: 2.7.5.0
Module: Tims-MaxDIA
Organism: Drosophila
Library: predicted
Dataset / acquisition
Samples: 8
Fractions per sample: 3
Total raw files: 24
Gradient: 2 hours per fraction
Window design: optimized variable windows generated with py_diAID
MaxQuant settings
Threads: 24
Deafult settings, specifically:
Enzyme: Trypsin
Max missed cleavages (DIA): 1
Default modifications and peptide length settings
No other changes from defaults (happy to share the mqpar.xml if helpful)
Observed runtime
Total processing time was ~18 days.
From runningTimes.txt, the dominant steps were:
DIA library search3: 20,957 min (~14.5 days)
Deisotoping: 1,865 min (~31 hours)
Create slice data: 1,151 min (~19 hours)
Hardware / environment
OS: Windows Server 2019
CPU: 48 cores
RAM: 512 GB
Storage (main run): 22 TB SAS HDD (7.2K RPM, 12 Gbps interface)
Because the runtime was so long, I suspected disk I/O. I repeated a subset run (2 samples) on a boot-optimized M.2 SSD (~480 GB) and compared it to HDD. Both took ~7 days for the same subset, so SSD vs HDD did not materially change the runtime.
Is it expected that “DIA library search3” dominates runtime to this extent for predicted libraries / TIMS DIA?
Given SSD did not improve runtime, what are the likely bottlenecks (CPU, RAM)?
Any recommendations on what to upgrade to reduce runtime if we want to run Tims-MaxDIA routinely?