sjdirs?

39 views
Skip to first unread message

Swethaa NG

unread,
Aug 12, 2024, 9:12:43 AMAug 12
to Biociphers
Hello again,

Hope you are doing well :)
I have a naive doubt about sjdir. The documentation reads " sjdirs: Comma separated list of paths where the sj files from previous builder runs are located. Used when running majiq with --incremental" I do not understand what does the previous builder runs mean?

I have SJ.out.tab files from STAR mapping runs, plus I also have splice junctions generated from regtools. There seems to be some discrepency in them, like the coordinates do not seem to match. I am still trying to understand that part, but can you please elaborate on this feature? 

Also, is the SJ.out.tab file from STAR enough for the builder run or is there any splice junction formatting needed to be made?

Please let me know, thanks a lot!
Kind regards,
Swethaa

Swethaa NG

unread,
Aug 13, 2024, 11:09:05 AMAug 13
to Biociphers
Hi, I am really really sorry for the spam. Please excuse. 
The reason I would like to know about SJ feature is because I I think I might be doing something wrong here, and this is causing some discrepency in my final results.

I mentioned the file paths of the SJ files (generated from STAR) in the builder config file and ran the pipeline (with deltapsi followed by voila modulizer) for one sample in each condition (condition 1: control , condition 2: patient). When I checked the modulizer summary tsv for the LSV coordinates, and out of curiosity checked them in IGV, I do not happen to see the LSV events like casette, alt5 etc.. even though they show a low prob_changing and high_prob_non_changing values..
I am yet to use voila view but I was hoping I should see the events in  IGV, right? Please let me know, thanks!

Kind regards,
Swethaa

San Jewell

unread,
Aug 13, 2024, 4:19:58 PMAug 13
to Biociphers
Hi Swethaa,

In the future, for a different question topic, I would recommend that you open a separate thread on the board. This makes the question easier to search for other users.

For your first post:

The sj files are only majiq-generated SJ files, not other software. Basically, the build will take all specified experiments in the build config file and make a unified splicegraph. When doing this, a large part of the runtime comes from the process of converting the BAM file into a series of splice junction spanning reads. Sometimes, you might want to re run a build and include new experiments. So, instead of requiring you to re-process the BAM files every time, we specify an intermediate format .sj which will be stored in the .sjdirs location(s). This means, when running the builder with --incremental , the program will not re-process bam files when an SJ file was created previously.

For your second post:

Please note I do not have a good handle on what data is available in IGV. I do not know if de-novo patient specific LSVs would be in their database or not. This would probably require more information about your experiment. Can you give an example of an event you see that is not matching with what you expect?

Thanks,
-San

Swethaa NG

unread,
Aug 25, 2024, 1:55:53 PMAug 25
to Biociphers
Hi, 
Hope you are doing well,
Sorry for the mixed doubts and thanks a lot for the clarification! I understand better about .sj files now!!
And I may not have understood some details in modulizer, I shall ask them in a separate post no worries. :)
Thanks!
Swethaa
Reply all
Reply to author
Forward
0 new messages