When browsing the Developmental Transcriptome “Gene Level RPKM” viewer, I noticed that the expression values (in log RPKM) in the heat map do not match the values for the same gene (same entrez number and name) in the “expression_matrix.csv” file for either the RNA-Seq Gencode v10 summarized to genes or the RNA-Seq Gencode v10 summarized to exons. However, when I click “Downlaod this data” under the heat map, the numbers in the heat map do indeed match what is in the “Expression.csv” file for that heatmap.
Can you help me understand why the values in the are different? I was under the impresson that the heatmap and the RNA-Seq Gencode v10 contained the same RNA expression data from the same samples.
Also, I noticed the values marked as N/A in the heat map are listed as 0 in the “Expression.csv” file for that heatmap. Just want to confirm whether N/A means not applicable/available OR means none measure as in 0 quantity as displayed in the csv file?
Thanks,
Matt
Hi @mshtrahman ,
As you’ve noted, in all heatmaps, the downloaded values match exactly what is shown on the screen. As for why the values don’t match, they should, so I’m not sure. Unfortunately the data for this project was done a long time ago from by a collaborator group so I don’t know if we’ll be able to track down an answer, but I’ll try and will post an update if I learn more.
One possibility is that the heatmap is still showing the Gencode v3c data rather than the newer Gencode v10 data (which could be checked). Another is that the heatmap is using a slightly different normalization, but that seems unlikely.
As for the N/A vs. 0 question, N/A means 0. This only shows up in genes with low expression and all samples have data collected for the exact same set of genes.
This post might provide some additional context.
Best,
Jeremy
Thanks Jeremy. That’s what I needed to know. I’m really just interested in having a searchable database of normalized RNA expression values for as many genes as I can find in normal human brain samples, so if RNA-Seq Gencode v10 summarized to genes contains that information and I can trust the values, then I guess it doesn’t really matter to me if it matches exactly what is in the heat maps. That being said, if you or someone else does figure out why there is a discrepancy, I would love to hear the answer. Unitl then, we will just use RNA-Seq Gencode v10 summarized to genes and trust that it is correct.
Thank you!
Matt
Yes, you can trust the RNA-Seq Gencode v10 summarized to genes and related files
