search

hakyimlab / PredictDB-Tutorial

Tutorial for running the PredictDB pipeline with GEUVADIS data
9 stars 11 forks source link

transformed_expression.txt? #4

Open jcabanad opened 3 years ago

jcabanad commented 3 years ago

Hi,

I'm using your tutorial and I have a doubt regarding this sentence:

"We want to generate PEER factors https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3398141/ and use those as our covariates and then perform a multiple linear regression for each gene in our gene expression matrix and then save the residuals from the regressions as our new expressions."

After seeing this sentence, It seems to me that the "transformed_expression.txt" file that we are going to use later in the script will be the our expression data modified according PEER factors, that in the script is named "residuals_expression.txt". But it is not like this, according the script, "transformed_expression.txt" is just our expression data transposed (the same than we used to built PEER factors).

It would be like this or there is an error in the script? Why do we need this file "residuals_expression.txt"?

Additionally, I'm having a lot of troubles installing PEER tools. Which version would you recommend me to use R or Pyhton?

Thank you very much.

Best regards,

Judit

Fnyasimi commented 3 years ago

@jcabanad The adjusting for the covariates (PEERS in this case) happens within the elastic net script. The residual expression was made available if you need to do other analyses apart from the model training. Additionally you can combine other covariates together with peer factors.

For PEER tools installation follow this guideline