Why some Change happen in 'ImmuneSigs_function' in ImmuneSigs68_function.R?

[huangwb8]

Recently I'm trying to learn how to use the pipeline in 'How_to_produce_gene_set_scores.ipynb' in this project. I was surprised when I run ImmuneSigs_function and get different result from the example, based on the same dataset. I check the code of ImmuneSigs_function in ImmuneSigs68_function.R, and I found something strange. I believe that the result changes as the code changes at the same time. First, the parameter of "doSig" is intentionally set by the author afterwards, which limited the calculation of most subtype of immune signatures. Second,some calculation code are unavailable by putting "#" in front of them. So, my question is : why do you have to do this? Is there any resonable reason? Looking forward your return~Thanks!

[Gibbsdavidl]

Hello. It looks like you've found the R code. :-)

First of all, if you want to call immune subtypes, I would encourage you to use this: https://github.com/Gibbsdavidl/ImmuneSubtypeClassifier

We found that the code used in the PanCancer project didn't work well with gene expression data from other software pipelines. The new R package addresses this problem.

But, if you want to see how the original immune subtypes were called, then you're in the right place. Now, a large portion of this code was given to me by biostatisticians, so I didn't write it, but I think I know what's going on.

OK, so you're at the part where the data has been transformed, as shown in the notebook. And you're calling scores <- ImmuneSigs_function(transformed_data, others, ...)

What are the scores? They don't match what's given?

In the ImmuneSigs68_function.R file, I don't see anything called 'doSig' ? So, can you ask the question again?

For your second question, I imagine that when Denise, who developed this code, was writing this particular script, she may have had some code in there for testing, but then commented it out because it was not needed. It's very common.

I guess I don't understand the question.

-dave

[huangwb8]

Thanks for your quick return! Amazing~ I think I had found the original reason, and it seems like a trick from myself(～￣▽￣)～ Because there are two so-called " ImmuneSigs_function" in this project~ One is in .\Immune-Subtype-Clustering-master\Scripts, and another one is in .\Immune-Subtype-Clustering-master\shiny-app\Immune-Subtype-Clustering\src . Although they share the same name,actually the latter is adjusted to only calculate 5 immune usefull signatures according to the research. And I had no idea why I choose the deeper code first... o(￣▽￣)ブ Anyway, it's not a big deal now. Problem solved! And I'll learn the new R package ImmuneSubtypeClassifier~ Thanks again~