Working on TCGA data using mixOmics

Analysis
1

Dear MixOmics team,

Many thanks for developing mixOmics tools/package that help us a lot to apply different data mining approaches on RNAseq data. I am very interested to do the analysis like what you did with breast cancer from TCGA and expand it a bit on another cancer type.
In your work flow you mentioned that the data is downloaded from TCGA and did not tell (as far as I checked) how you divided them to data.train and data.test (I also checked the diablo paper S2 section). I did downloaded the data but do you have any scripts that can help me to proceed with further analysis/normalization and make the same train data set for another cancer types?

Many many thanks in advance,
Kind Regards,

2

The protein data served as the splitting factor, as it had the least number of samples available. The training data comprised of 4 data-types (mRNA, miRNA, CpGs and proteins), whereas the test data did not have any proteomics data (see Table 1 in this paper). Since the predictions are made separately for each dataset (block) and compiled using average or majority vote (for a consensus prediction), DIABLO allows for entire datasets to be missing in the test set.

you can find the links to the data compilation and manuscript analysis repos here.

1 Like