Table 1 Results of the cross-validation and independent set tests. A. Correct classification rate results obtained from the three main G-protein coupling groups, in a five-fold cross-validation procedure. The training set was randomly divided to five equally balanced sets. Afterwards, we trained a model using the sequences in the four sets whereas the last set was used for testing. This procedure was repeated five times. B. The library of refined profile Hidden Markov models (pHMMs) derived from the primary dataset of 282 GPCRs (see text) was tested against a validation set comprised of all GPCR sequences of subtypes with known coupling preference summarized in [37], excluding the sequences used to train the models (479 GPCRs in total). This independent test yielded 91% correct classification rate. Numbers in the diagonal of the charts represent true positive predictions. The total number of predictions for each group (row) is not equal to the total number of observations, since several GPCRs were not classified in any group.

Five-fold cross-validation test

G _i/o

G _q/11

G _s

total

observed

G _i/o

109(90.8%)

1

1

120

G _q/11

2

78(82.9%)

2

94

G _s

0

0

66(97.1%)

68

111

79

69

253(89.7%)

B

predicted

Validation test (479 GPCRs)

G _i/o

G _q/11

G _s

total

observed

G _i/o

233(91.4%)

16

4

256

G _q/11

9

90(88.2%)

2

102

G _s

6

2

113(93.4%)

121

248

108

119

436(91.0%)