Recently, I'm reading a paper and checked the attachment(https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE159919). I found that read counts for gene ENSMUSG00000064356 in file GSE1599_PolyA_read_counts is 0 in sample mES_Chr_rep_1_polyA. While the TPM of this gene in file GSE159919_polyA_gene_expresion is 6289.54. That's confusing. Then I asked the author of this paper. He/She answered, "Both are correct due to the fundamental difference between feature count and Kallisto.". Can someone tell me is that possible? The feature count is 0 while TPM is more than 5,000. I checked the calculation method of TPM, and if the feature count is zero then the TPM should also be 0.