Four weeks after Erich Wanker?s team published its human interactome paper, Marc Vidal?s group at the Dana-Farber Cancer Institute in Boston has followed suit.

In today?s (Sept. 28) online edition of Nature Vidal?s team reports its analysis of interactions between 8,100 or so human open reading frames corresponding to some 7,200 protein-coding genes. From that ?search space? of 7,200 x 7,200, or nearly 52 million combinations, the group used a high-throughput yeast two-hybrid approach to identify 2,754 interactions among 1,549 proteins, which they calculate represents ?about 1% of the human interactome? (not including alternate splicing variants).

As with Wanker?s paper, Vidal?s human interactome is a technical tour de force. What I find most striking is the length to which Vidal?s team went to validate their data. Random verification using co-affinity purification suggests nearly 80% of the data are technically correct (that is, are not ?technical false positives?) The group then worked on the biological false positives, cross-checking its interactions against coexpression data, shared Gene Ontology annotations, shared upstream regulatory elements, and shared knockout phenotypes in mice.

Between the two studies, researchers have some 6,000 new interactions to work on (not counting overlap). If Vidal?s calculation is correct, and even if the two datasets share no interactions in common (not likely), that still represents only a tiny fraction of the total human interactome.

Look for Johns Hopkins University?s Joel Bader to add another drop to the bucket in the months ahead.