Skip to main content

Table 6 Statistical significance test for global quality prediction

From: Improved model quality assessment using ProQ2

Method   1 2 3 4 5 6 7 8
ProQ2 1   -14.48 -21.73 -56.80 -39.07 -47.35 n/a -103.07
QMEAN 2 -29.86   −1.91 -43.51 -18.50 -30.30 n/a -106.70
MetaMQAP 3 n/a n/a   -38.46 -11.29 -23.67 n/a -107.76
ProQ 4 -61.20 -39.03 n/a   -20.91 -8.69 n/a -114.42
ConQuass 5 n/a n/a n/a n/a   -4.79 n/a -110.58
Distill_NNPIF 6 n/a n/a n/a n/a n/a   n/a -112.22
MULTICOM-CMFR 7 -49.09 -17.94 n/a -14.99 n/a n/a   n/a
QMEANclust 8 -65.82 -79.03 n/a -91.56 n/a n/a -86.12  
  1. Pairwise statistical significance test on the correlation coefficients for global quality prediction from CASP8 (below diagonal) and CASP9 (above diagonal). The values correspond to the logarithm of the P-value for methods being statistically indistinguishable, obtained by comparing the distribution of Fisher’s Z (Eq: 1) for the correlation coefficients, R, from Table 5. Values in italics are not distinguishable at the 10−3significance level.