meyer-lab / mechanismEncoder

Developing patient-specific phosphoproteomic models using mechanistic autoencoders
4 stars 1 forks source link

Patient CPTAC data #26

Open aarmey opened 3 years ago

aarmey commented 3 years ago

Is there somewhere we can find this? Even if there's some normalization correction to be performed, is there a table setup with this data we can import from? @JacksonLChin has a standard autoencoder setup and would like to start with some of the necessary architecture experiments.

sgosline commented 3 years ago

Hi all,

I just uploaded most of the un-normalized AML patient data to synapse:

These replace syn22156830 and syn22314122 respectively.

sgosline commented 3 years ago

As for the other tumor types, we can use the cptac module that I committed to the repo. However, as far as I know it's unclear whether or not normalization has been done. I think it has been done for ovarian but not for others. I think the best way to figure this out a posteriori is to do a correlation between global levels and phospho sites, assuming there will be more correlation in the case where there was no normalization.

aarmey commented 3 years ago

Ok, thanks Sara!

sgosline commented 3 years ago

Update: there are actually three synapse tables that contain uncorrected phospho data. I realized my code is written in R with packages, so it'd be better just to list them here (@JacksonLChin ):