References | Gavin et al. [7] | Ho et al. [8] | Krogan et al. [9] | Gavin et al. [10] | Krogan et al. [11] |
---|
[7] | - | 82 | 51 | 442 | 334 |
[8] | 516 | - | 25 | 222 | 286 |
[9] | 299 | 246 | - | 121 | 151 |
[10] | 1,143 | 717 | 371 | - | 1,128 |
[11] | 1,149 | 1,277 | 478 | 1,732 | - |
- As in Table 2 the values above the diagonal give the number of viable baits in common between each pair of experiments, and the values below the diagonal give the number of viable prey in common. Again, the overlap is very small. Consider the two largest experiments carried out so far: with a set of 2,264 viable baits and 5,323 viable prey, Krogan et al. [11] tested for the presence of at least 12 million complex membership interactions. Gavin et al. [10], with 1,752 viable baits and 1,790 viable prey, tested for at least 3.1 million interactions. However, even for these two datasets, the largest so far, the known overlap is only 1,128 × 1,732 ≈ 2.0 million. One of the possible explanations for these low estimates of coverage and overlap is that our definitions of viable baits and viable prey are restrictive and that indeed a much larger space of interactions might have been tested. For example, Gilchrist et al. [29] estimated a value about twice ours for the number of tested prey in [7]. This situation will hopefully be alleviated as researchers report more complete data on which interactions were actually tested.