Table 6 Clustering performance of MCODE, MCL and MRF: comparison with the MIPS annotations. We use all proteins in the experiment with annotation.
From: Identifying protein complexes directly from high-throughput TAP data with Markov random fields
Dataset | MCODE | MCL | MCL with inter. prob. | MRF (spoke) | MRF (matrix) | |
|---|---|---|---|---|---|---|
Gavin02 | CO | 29.0 | 61.5 | 62.6 | 64.4 | 66.4 |
PPV | 73.6 | 71.3 | 71.7 | 73.5 | 66.9 | |
Acc | 46.2 | 66.2 | 67.0 | 68.8 | 66.6 | |
All pairs | ||||||
SN | 2.3 | 68.6 | 68.9 | 66.7 | 62.6 | |
SP | 92.5 | 78.7 | 82.4 | 87.9 | 64.7 | |
Geo. average | 14.7 | 73.0 | 75.4 | 76.6 | 63.6 | |
Gavin06 | CO | 33.7 | 64.0 | 65.7 | 66.0 | 67.7 |
PPV | 79.0 | 62.6 | 68.6 | 70.4 | 67.3 | |
Acc | 51.6 | 63.3 | 67.2 | 68.2 | 67.5 | |
All pairs | ||||||
SN | 4.9 | 44.1 | 44.7 | 37.2 | 38.2 | |
SP | 79.6 | 18.0 | 22.5 | 70.0 | 66.1 | |
Geo. average | 19.7 | 28.2 | 31.7 | 51.0 | 50.2 |