I've taken this out of the comment section, as I'm trying to provide an answer, so here goes:
No offense, but I think you still want to take a little bit of time to consider what you really want to show in your heatmap, and where those number would come from.
For instance, in your scenario: if you want to show differences in expression between a few (two) groups (organs, for you), you probably want to show the relative expression of the genes of interest from all of your samples, and not not a two column matrix.
Consider googling around a bit for heatmaps, gene expression and R, or find one of your favorite publications from recent memory that have a heatmap figure you particularly liked -- what are they showing?
You should also read through some of the RNA-seq workflows, which also have their own expression-based heatmap section, such as:
- RNA-seq ... with limma, Glimma, and edgeR
- edgeR QLF
These begin from different starting (data) points, but the heatmap is constructed from some sort of normalized count data, which correspond to the RPKM values you have, which has been log transformed.
Once you understand what is being plotted in those vignettes, you might just go with the tools they used there, or learn how to use ComplexHeatmap and really up your game -- it's really a remarkable package with incredible documentation.