Hi,
Many researchers have found that clustering transcriptomes by similarity often leads to a pattern where the samples cluster by species rather than by homologous cell types or tissue types, even when the cell types are much older than the species compared (aka "species signal").
The current consensus seems to be that this is caused by a batch effect, essentially an artifact of experimental design. But this explanation is unlikely in at least some of the data we analyzed.
In a recent paper posted on BioArchives my colleague Jake Musser and I propose an alternative interpretation:
[URL="http://biorxiv.org/content/early/2015/05/16/019380"]
a species signal naturally arises when the transcriptomes of different cell types in the same species undergo concerted evolution, i.e. if mutations lead to simultaneous changes of the expression status of a gene (essentially gene expression pleiotropy). A more technical paper is in preparation.
Many researchers have found that clustering transcriptomes by similarity often leads to a pattern where the samples cluster by species rather than by homologous cell types or tissue types, even when the cell types are much older than the species compared (aka "species signal").
The current consensus seems to be that this is caused by a batch effect, essentially an artifact of experimental design. But this explanation is unlikely in at least some of the data we analyzed.
In a recent paper posted on BioArchives my colleague Jake Musser and I propose an alternative interpretation:
[URL="http://biorxiv.org/content/early/2015/05/16/019380"]
a species signal naturally arises when the transcriptomes of different cell types in the same species undergo concerted evolution, i.e. if mutations lead to simultaneous changes of the expression status of a gene (essentially gene expression pleiotropy). A more technical paper is in preparation.