How to binarize genomic data

Question

[mirko.hu]

I have a table of genomic data. The rows are the patients, the columns are the genes and the entries are the gene abundances. I need to binarize the data so that above a threshold the entries are substituted by 1's and below the threshold, the entries are substituted by 0's. Which is the best way to do this? How do I find the correct threshold?

I need to binarize the data to obtain the correspondent biadjacency matrix of an unweighted network. Thank you for your help.

Answer 1

[Mensur Dlakic]

One way to do it is with sklearn. Don't know if that's the best way.

Answer 2

[venu]

How do I find the correct threshold?

This can vary for each patient, depending on the heterogeneity in your cohort. IMO, it's hard to find one threshold for all samples/genes. What you can do is

• add quartiles to all genes in each patient
• Take 3rd and 4th quartiles as 1 and rest 0 for each patient

You can do it basic R/python skills.

P.S: Not sure what your goals are, the answer is just to let you know how to do it. How much sense it makes sense biologically is up to you.