Psix on 10x data runs but saves nothing to directory "save_files_in"
Closed this issue · 2 comments
Hi Carlos,
I have 10x (UMI) data which I processed using STARsolo to get psix_annotation.tab.gz and I've created the intron_junction annotations as in #3 from the same genome assembly I used for STARsolo. When I run the following code, psix runs for ~4h and finishes without an error. But the folder psix_output/ will still be empty (I'd at least expect psi.tab.gz in this folder, right?). Is combining tenX=True and solo=True even supported?
import psix
psix_object = psix.Psix()
psix_object.junctions2psi(
sj_dir='/count/Solo.out/SJ/raw/',
intron_file='psix_annotation.tab.gz',
save_files_in='psix_output/',
tenX = True,
solo = True
)
The annotation file looks like this (first 2 of 144493 rows):
| index | intron | event | gene |
|---|---|---|---|
| Synrg_1_I1 | tig00000049:760716-763023:+ | Synrg_1 | Synrg |
| Synrg_1_I2 | tig00000049:763327-764633:+ | Synrg_1 | Synrg |
The SJ output from STARsolo also looks good (mtx + 2 tables for barcode and feature, all non-empty).
Many thanks in advance for your help.
Best,
Stefan
Oh, there was a bug on junctions2psi that did not save the PSI and mRNA files when solo = True was passed. This was completely unintentional. The issue should be resolved now (check the updated notebook for the midbrain example, and the directory where the files are saved).
Combining solo = True and TenX = True is supported.
Please make sure to update Psix; and I do apologize for the oversight.
By the way: I have only done some limited testing of Psix on 10X data, which is why the Readme indicates it's on Beta. I'm still experimenting with the parameters on 10X data, but I think that lowering some filtering requirements (such as setting min_observed = 0.1) might increase the sensitivity, since the data is sparser. Please don't hesitate to let me know if you run into another problem.
That fixed it. Now it runs perfectly and produces meaningful output.
Thanks so much for your quick help!