majiq heterogen grouping problem： Should different tumors or subtypes be run separately?

[qk Wang]

When I perform Majiq heterogen on three tumors and normal samples, I cannot distinguish the three tumors after dimensionality reduction. However, when I run the tumors separately with normal tissues, they can be separated. However, different subtypes of the same tumor cannot be separated. Do I need to compare the different subtypes with normal tissues separately before merging them? Does Majiq blur the differences within the group?

Also, my psi file has a few samples with very high -1 values. I checked the mass mapped percentage of bam and it is normal. Is this normal? Hi everyone, I hope to get a reply. Thank you so much

[San Jewell]

Hello,

I appreciate your question. I think to give you the most accurate answer, I will need a little  more information on some of the terms you use.

1) You mention two types of runs you do, and results that you see. Can you articulate this? When you run separate or together what does this mean? Different builds completely? Different groups in the build? Different groupings when running "majiq heterogen"? It might help if you post a redacted version of the commands you use and your build config file.

2) When you see some results that "can be separated" and "can not be separated", what do you mean by this? What are you seeing in one that you are not seeing in others? Certain junctions/lsvs? A screenshot or two might help me understand.

3) You mention "very high -1 values". What does this mean? Like -0.9? Is it PSI or dPSI? What is the column and what are some of the values you see?

Thanks,

-San