Additional file 5: Table S1. of Tissue-aware RNA-Seq processing and normalization for heterogeneous and sparse data

Breakdown of gene types remaining in each data set after different filtering approaches. Filtering in a tissue-specific manner, we keep genes that appear in a least half the number of samples present in of the smallest phenotype group (for GTEx, at least 18 samples since the â smallestâ tissue has 36 total samples); this leaves 30,333 genes of which 60% (18,328) are classified as protein coding genes and 11% (3220) are pseudogenes. This contrasts with our tissue-agnostic method in which genes are removed if they appear in fewer than half of the samples in the data set; this retains only 15,480 genes for which 84% (12,994) are protein coding, and 4% (659) are pseudogenes. (XLSX 36 kb)