TABLE 2.
Correlation of structure rank in different models
| HP′ (6693) | HPmax (1224) | AB (4127) | AB′ (4490) | ABmax(1577) | |
|---|---|---|---|---|---|
| HP (1475) | 0.74 (1317) | 0.56 (331) | −0.15 (550) | −0.17 (540) | 0.10 (327) |
| 89.3% | 27.0% | 37.3% | 36.6% | 22.2% | |
| HP′ (6693) | 0.59 (310) | −0.19 (964) | −0.20 (1004) | 0.11 (306) | |
| 25.3% | 23.4% | 22.4% | 19.4% | ||
| HPmax (1224) | 0.20 (1106) | 0.20 (1106) | 0.18 (1172) | ||
| 90.4% | 90.4% | 95.8% | |||
| AB (4127) | 0.98 (4095) | 0.92 (1493) | |||
| 99.2% | 94.7% | ||||
| AB′ (4490) | 0.91 (1493) | ||||
| 94.7% |
Structures are ranked by the sizes of their neutral sets in a given model; the structure with the largest neutral set is ranked 1, and so on. When two or more structures have the same neutral set size, they are assigned the same rank, in a “golf ranking” manner. For example, in the HP model, the three largest neutral set sizes are 48, 37, and 36, and the number of neutral sets with such sizes are, respectively, 2, 4, and 2. Thus, the structures encoded by these eight neutral sets are ranked accordingly as 1, 1, 3, 3, 3, 3, 7, 7. The rank of a structure depends on the model interaction scheme. For every pairing of the six models studied here, the Pearson correlation coefficient for the rank of the (overlapping) structures that are encodable in both models is provided as the first entry in each box of this table. The number of overlapping structures is given in parentheses (second entry in each box). This overlapping number as a percentage of the total number of encodable structures of the less encodable model is also provided (bottom entry in each box). The total number of encodable structures for each of the models (from Table 1) is shown in parentheses with the model names.