I have a count matrix of gene expression data for several samples, derived from heart tissue (pig). I want to confirm that the gene expression pattern in these samples is consistent with heart tissue.
Is there any database, or other tool, to find genes that are only expressed in this given tissue? Negative markers would be good too, in analogy with the use of positive and negative surface markers in immunohistology. The expected result is to say: "genes x,y are expressed in these samples, therefore the samples are likely derived from heart tissue."
I have checked RefEx database. When I look at heart tissue I get a list of genes, but none of them seems to be specific for heart, or else I am misinterpreting the results.
NOTE: I am aware that databases for pig might not be available, but starting with human and finding homologs is a viable option.
A great tool for this would be CellNet from Patrick Cahan's lab, it can classify your samples into a variety of tissues based on gene regulatory networks. But it's only available for human and mouse data so far. However, it includes tools for it to be trained for other species. I understand it may not be feasible for you to use it because getting a good training dataset and doing the training could require a substantial effort you might not be able to justify. But if you had mouse or human data in the future that you'd wanna do the same thing with, this would be a cool way to go. Links to Nature Protocols (2017) and Cell (2014) papers: http://dx.doi.org/10.1038/nprot.2017.022 http://dx.doi.org/10.1016/j.cell.2014.07.020
The UCSC genome browser has expression information for human genes, maybe you can download it using table browser and use the data to find the most likely tissue.
Checked it out. Going through UCSC genome browser means I have to look up every gene manually and see if any gene is uniquely expressed in heart?
No, you can download all the data using the Table Browser (tools->table browser)