6
$\begingroup$

I am looking the secretome profile and the membrane receptor profile for a given cell type.

In my specific case, this should be the secretome and outer membrane receptor profiles of dorsal root ganglion.

What I've done in the past is taken proteomics datasets or RNAseq datasets and used this as a reference cell surface proteome or secretome.

I wasn't sure what the best way to filter secreted proteins or membrane receptors were, so I used the Human Protein Atlas and this turned out to be a big headache. A possible alternative method might be to use DAVID, but there must be a more efficient method via the command line.

My questions are these:

  • Is there a "consensus" database containing protein expression profiles for different tissue types in human?
  • If not, what is the best way to filter out the secreted and transmembrane proteins by gene symbol?
$\endgroup$
1
$\begingroup$

I don't think there is a database with expression data for your particular cell type. I am afraid your "headache" approach doesn't sound so bad...

So get expression data e.g., from GEO for your cell type, and define which genes/proteins you consider as "expressed" (e.g., minimal RPKM value).

Then, you can use GO annotation to find your genes involved with membrane or secretion (or what ever). You can use biomaRt in R for this.

For example, you want to know all "membrane" genes.

> library(biomaRt)
> 
> ensembl = useMart("ensembl",dataset="hsapiens_gene_ensembl")
> 
> membrane.genes <- getBM(attributes=c('hgnc_symbol',
> 'ensembl_transcript_id', 'go_id'),
> filters = 'go', values = 'GO:0016020', mart = ensembl)

And now you want to know it only from genes that are expressed in your ganglion cells (with dummy example of two genes).

> gene.list <- c("KIR2DL3", "GPR1")
> 
> membrane.genes.list <- membrane.genes[membrane.genes$hgnc_symbol ==
> gene.list,]
> 
> membrane.only <- membrane.genes.list[membrane.genes.list$go_id ==
> 'GO:0016020',]
$\endgroup$
3
  • 1
    $\begingroup$ Note that this will only retrieve genes directly annotated to GO:0016020 (membrane). For example, the gene PLSCR5 is not in the list despite this being annotated to plasma membrane (GO:0005886). Also, PM may be a better choice for exploration of cell-type specific surface receptors. $\endgroup$
    – cmungall
    Aug 14 '17 at 0:29
  • $\begingroup$ I try to give an example of how to use biomaRt for GO annotation. GO annotation is not a holy grail, but just a tool that can be helpful in your research sometimes. $\endgroup$
    – benn
    Aug 14 '17 at 7:49
  • $\begingroup$ Agreed, it's a good answer, just pointing out this for other people coming by this question and using biomaRt for other GO queries. $\endgroup$
    – cmungall
    Aug 14 '17 at 14:47
1
$\begingroup$

Is there a "consensus" database containing protein expression profiles for different tissue types in human?

https://www.ebi.ac.uk/gxa/home

If not, what is the best way to filter out the secreted and transmembrane proteins by gene symbol?

Personally I like to run a local copy of SignalP, as it will yield consistent results. However bioinformatic approaches tend not to capture highly unusual transmembrane proteins or membrane associations (e.g.: only one leaflet etc.). http://www.cbs.dtu.dk/services/SignalP/

Depending on the organisms, and scope of your analysis, further possibility could be to visit supplemental tables of recent publications containing somewhat supervised computational predictions on secretion, e.g.: http://science.sciencemag.org/content/356/6340/eaal3321

I am looking the secretome profile and the membrane receptor profile for a given cell type.

The above links don't inform about cell-type specific retention of potentially secreted proteins. - but their combination will yield cell-type specific RNA and protein levels of transmembrane and secreted proteins.

$\endgroup$
0
$\begingroup$

You can get expression data from databases and for membrane proteins, run a transmembrane protein prediction method like CCTOP (http://cctop.enzim.ttk.mta.hu/).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.