Normalising heatmap counts

[4galaxy7]

I'm trying to produce a heatmap for my RNAseq dataset (using RPKM values). The counts vary from 5 - 40 000+. Obviously when I produce the heatmap, using Pheatmap, the values are very skewed towards the top end (see photo).

Can anyone reccomend a good method of normalising / transforming the data to give a better visual representation?

Thanks

[wdecoster]

Most commonly log scaling is used. Recommended is probably to use DESeq2 or similar to extract log normalized counts with a more sophisticated model than RPKM.