slsevilla
commented
2 years ago Testing params:
- used test_2 for all tests (1 multiplexed sample, 2 project samples)
Documentation:
- iCount demultiplex: https://icount.readthedocs.io/en/latest/_modules/iCount/demultiplex.html
- Ultraplex: https://github.com/ulelab/ultraplex
Reasoning:
- Demultiplexing was taking a significant amount of time per sample. Seeking a comparable tool that would allow for an increase in speed.
Results:
- Time: Ultraplex significantly reduced the run-time for this test set.
| iCOUNT | ULTRAPLEX | |
|---|---|---|
| Time | 25.5 hours | 1.15 hours |
- Barcodes: No significant difference was noted between methods, where the first image represents iCount and the second Ultraplex.
ultraplex is now installed on biowulf as a module ... did some testing.. it was able to demultiplex, quality trim and remove adapter (all in one step) for the 520+ milllion read fastq file (FlagRo1hr_S1_R1_001.fastq.gz from latest Sohyoung sample) in < 20 minutes with 56 core node. Chill, Samantha (NIH/NCI) [C] Check /data/RBL_NCI/Wolin/mESC_clip_4/manifests/ultraplex_testing/FlagRo1hr_ultraplex_output for details. the script run_ultraplex.sh has all the paramters I used.... we need to get this in the pipeline ASAP!!! These are huge savings.... No more Scatter + Gather.