Nature Methods | Methagora

Searching high and low for protein-protein interactions

If you were given a choice between a large protein-protein interaction data set obtained by high-throughput experiments and another data set extracted from a collection of small-scale experiments via literature mining, which one would you trust more?

We think that high-throughput approaches to protein-protein interaction study have made constant progress over the past few years. In the May Editorial, we explain why we think it is now time to abandon the idea that the body of information on protein interactions available in the literature is superior to results of high-throughput experiments.

What about you: When it comes to protein-protein interaction studies, are you a high-throughput friend or foe?


  1. Report this comment

    chenfeng said:

    I think both of them should be well exploited.

  2. Report this comment

    Peter Uetz said:

    After having studied one protein during my PhD for 4 years, I eventually found a hand full of interactions. So I decided to go high-throughput (HT). HT screening is not a matter of preference, it is a matter of necessity. There is no reason why high-throughput data should be of lesser quality than small-scale data (especially if HT data is followed up and that is the whole purpose of it).

  3. Report this comment

    Thomas Joseph said:

    Given the choice – I would choose the data set extracted from a collection of small-scale experiments via literature mining. But both sets of information need to be curated.

  4. Report this comment

    David Scott said:

    It really depends on what information you want. Do you want (a) a yes/no answer (b) a dissociation constant or © the full thermodynamic profile in terms of enthalpy, entropy, free energy, heat capacity changes and changes in accessible surface area. HTP technology is very good at (a), has some passable success at (b) and absolutely nowhere with ©. So, depending on the model you wish to build up and the detail you wish to include, depends on whether you mine the literature for all those sound biophysical studies, or trust what you get through HTP platforms.

    As ever, curation of the data and sound criteria to eliminate false positives/negatives is needed.