Human cell-surface protein databases or prediction algorithms

Question

[bw.]

I'm looking for an up-to-date resource on human cell-surface proteins - such as a database of all known human cell-surface markers, or a program that predicts trans-membrane domains based on protein name or AA sequence.

Ideally this resource would also be cross-linked to mass-spec or expression data by cell type.

The application is oncology cell therapy target identification.

Thanks

-Ben

Answer 1

[cdsouthan]

An old problem with no "clean" solutions but some approximations you might find useful. I am also intepreting some of your "or"s as "and"s

1. Selecting "(annotation:(type:signal) AND organism:"Homo sapiens [9606]") AND reviewed:yes" in UniProt gives 3,467, substituting keyword:"Cell membrane" gives 2871. The latter may be cleaner in Swiss-Prot than TrEMBL
2. You can explore intersects with the four mass-spec cross references (but PRIDE seems over-occupied) or the transcript expression sources but I dont know how to link these with cell types, and its likely to be noisy if you do (try BIOMART just for the intersects)
3. For oncology targets in general you should look at https://cansar.icr.ac.uk/
4. This is the type of data you realy want http://www.ncbi.nlm.nih.gov/pubmed/24555563 but I'm not sure how far off a selectable UniProt x-ref might be

[Shicheng Guo]

N=3610, annotation:(type:signal) AND reviewed:yes AND organism:"Homo sapiens (Human) [9606]"

Answer 2

[pld]

hmmer and blast will give you protein domains: http://hmmer.janelia.org

You could also check out http://www.immgen.org/, this would only work if you're interested in immune system cells however. This would be a potential source of cell-specific expression.

Several groups are working towards this, e.g. Cell ontology is trying to integrate GO and protein ontology, but like cdsouthan said, there's not a good solution just yet.

You could also use either a list of targets (gene/protein expression) or use GEO to pull gene expression data for relevant cell or tissue types and then filter the lists using gene ontology.