Contradictions between tissue-specific expression databases, which one to use?

Question

[hbkonuk]

Hi All, I'm trying to collect list of genes based on tissue-specific expression data. There are some tools containing tissue-specific genes for example, TIGER, PAGenBase, Unigene, GTEx and BioGPS (last two accessed via Genecards). But, there is no consensus among them. Here's results of UCN3 expression data accorrding to different databases:

• TIGER says UCN3 is specifically expressed in pancreas (link)
• PaGenBase shows us the same gene is specifically expressed in kidney (link)
• GTEx shows that UCN3 is expressed preferentially in colon, pancreas, pituitary and small intestine (link)

• The complicated output of BioGPS for UCN3 (link) shows that UCN3 is highly expressed in vastus lateralis muscle and missing in fallopian tube epithelium (if I'm not mistaken).

  Based on citation counts BioGPS has highest citations (645 citations) followed by GTEx with 347 citation. TIGER and PaGenBase have 147 and 16 citations, respectively.

Do you have any idea which database is more reliable? I am aware that TIGER is old and based on EST data, BioGPS is old and based on microarray data, on the other hand PaGenBase is newer and based on microarray and RNA-Seq data, GTEx is newer as well and based on RNA-Seq data (along with genotypes). In brief, how can I find reliable list of genes specifically (maybe prefertially) expressed in a specific tissue?

Actually, I have one more related question. Is there any tool or open-access resource available for cell-type specific genes (for example, genes expressed specifically or restricted in epithelial cells, dendritic cells etc.) How can I get cell type specific genes?

Thanks,

Hatice Büşra KONUK

Answer 1

[Charles Plessy]

The expression of this gene is quite low, so I would not interpret the differences between the databases as contradiction. Each of them has a different coverage of the tissue types, and uses a different technology, so lack of evidence for expression in one database is not a proof of no expression in general.

By the way, you may also be interested by the FANTOM5 expression database (link).

Answer 2

[Lars Juhl Jensen]

Similar to what Charles said, my interpretation would be that this gene is very lowly expressed. Not only is the expression so low that not having detected it in a tissue should not be interpreted as absence compared to other tissues. In fact it is appears to be so lowly expressed that I would question if it has even been reliably detected in any tissue.

My interpretation would thus not be that the data are contradictory. Rather, I would say that the datasets agree that we do not know where UCN3 is expressed, because its expression level is below the limit of what can be reliably called based on any of the studies.