Hi Hannan,
Thank you for your email.
The data_cna.txt files for TCGA PanCancer Atlas studies in cBioPortal represent GISTIC2.0 thresholded gene‑level copy number calls. These calls were generated from segmentation data derived from Affymetrix SNP6.0 array copy‑number profiles (in this case I believe CBS-based segmentation was used). The harmonized segmented data were then analyzed with GISTIC2.0 to identify recurrent amplifications and deletions across a cohort, and these results form the basis of the data_cna.txt thresholded calls you see in PanCancer Atlas and cBioPortal.
Here is a link with more information: