High-throughput human PPI dataset(s)

Question

[Louis]

Could anyone suggest genome-wide high-throughput protein-protein interaction screen done for human cells?

I can find two TAP high-throughput experiments (which seems to be the most trustworthy) but only for yeast.

I get the feeling this might be common knowledge to more seasoned bioinformaticists so please enlighten me if there are notable ones I should start off using.

Stelzl et al (2005) seems the best I can find at the moment, but I've read enough about Y2H false positives to be wary of using it to validate my dataset. I'm just after high-throughput experiments to try and eliminate the bias that arises in medical fields of more heavily studied proteins having more interactions mapped.

I've been reading up on IMex's component databases and am wondering if a better approach would be to somehow filter these datasets for interactions that had been found via high throughput experiments?

[russhh]

If the method you used was not a two-hybrid approach, I'd be more tempted to use Stelzl as your reference than a complex-pulldown experiment. The concern over false positives in Y2H always seemed a bit misguided.

[Louis]

The fact that Stelzl et al provide their PPI across 4 PDF supplementary tables has nudged me towards Mikhail's suggestion :-) Good to hear re: trustworthiness of the Y2H though

[russhh]

well, yes, that's how the data is provided within the paper. You could just obtain it from biogrid or similar, however.

Answer 1

[mikhail.shugay]

I recommend this paper. Its advantage is that it contains AA positions of protein interaction interfaces and corresponding Pfam domains (suppl table 3). They base on Y2H, but those are curated and well-verified interactions.