Hi all,
My technician asked me to compare the splicing events and splicing junctions for cancer and normal samples. I performed Tuxedo workflow for the RNAseq data. I have tophat, cufflinks, cuffmerge, cuffquant, and cuffdiff output files.
For example, I am interested in identifying differential alternative splicing in both the samples.
Is the below understanding correct?
Will I be able to get the below output from junctions.bed file from cancer and normal?
"E" refers to exons in the below scenario.
Gene ---------------------------------Isoform--------------------------Cancer----------------Normal
Gene1 (E1-E2-E3-E4-E5)-------E1-E2-E3-E4-E5--------------Absent----------------Present
Gene1 (E1-E2-E3-E4-E5)-------E1-E2-E3-E5-------------------Present---------------Absent
Gene1 (E1-E2-E3-E4-E5)-------E1-E2-E5------------------------Present---------------Absent
Gene2 (E6-E7-E8-E9)------------E6-E7-E8-E9-------------------Present --------------Absent
Gene2 (E6-E7-E8-E9)------------E6-E7-E9----------------------- Absent----------------Present
Gene2 (E6-E7-E8-E9)------------E6-E9----------------------------Present ---------------Absent
I haven't used cuffdiff in quite a while, I've historically found it to be to reliable to bother with. We usually use MATS for this sort of thing.
Thanks, Devon. I will try MATS.
Dear Devon,
Thanks for suggesting rMATS. I am done with rMATS for my data. I got the 5 output files (A3SS, A5SS, MXE, RI and SE) for each pair of cancer and normal samples. After performing some research on those output files. I understood that the below summary table was arrived by considering the FDR < 0.05 as significant entries under each of splice junction. As per the output, IJC_SAMPLE1 and SJC_SAMPLE1 is for cancer, IJC_SAMPLE2 and SJC_SAMPLE2 is for normal, IncLevel1 is for cancer, IncLevel2 is for normal and IncLevelDifference is the difference between IncLevel1-IncLevel2 .
The number present in IJC and SJC columns refers to the number of reads supporting that event.
When the IncLeveldiff value is positive they determine the cancer sample is significant, similarly, when the IncLevelDifference value is negative they determine the normal sample is significant.
How can I filter the splice junctions which are unique to cancer and unique to normal from those 5 output files?
You'll need to write a little post-processing script to handle this. Having said that, normally one puts all of the samples in the two groups in at once. I don't know if rMATs is able to use paired samples in this fashion, but if not perhaps it could be hacked to do so (I've never looked into the internals, so I don't know how relevant this is).
Tagging: Devon Ryan
Dear Mr. Charles Warden, Have you got a chance to work on the above scenario. Also, I want to filter the splice junctions which are unique to cancer and unique to normal from those 5 output files?