Table 2. Statistical evidence sorted by strength.
Amino acid |
PCA factor loadinga |
Raw correlationa |
Significance |
Raw Δ (S.D.) |
Δ (S.D.) |
Significance |
Statistic |
Gln (Q) | –90% | –80% | ∼10–8 | –2.18 (0.31) | –1.76 (0.25) | ∼10–4 | t-test |
Glu (E) | 80% | 80% | ∼10–6 | 2.27 (0.40) | 1.73 (0.31) | ∼10–4 | t-test |
Val (V) | 50 to 65% | 60% | ∼10–3 | 1.57 (0.42) | 1.40 (0.38) | ∼2 × 10–3 | t-test |
Thr (T) | –65% | 60% | ∼10–3 | –0.84 (0.25) | –1.31 (0.39) | ∼5 × 10–3 | t-testb |
His (H) | –40 to –60% | –60% | ∼10–3 | –0.44 (0.15) | –1.22 (0.42) | 1%b | t-testb |
Ser (S) | –30 to –60%b | –40% | 1% | –1.11 (0.51) | –1.18 (0.54) | 5% | t-test |
Asn (N) | –30 to –40% | –35% | 3% | –1.94 (n/a)b | –1.05 (n/a)b | <2% | Median/Mann–Whitneyb |
Arg (R) | 20 to 30%b | 25% | >5% | >0 (n/a) | >0 (n/a) | 1% | Mann–Whitney |
For each amino acid, the range of PCA factor loadings for component 2, the raw correlation to the binary variable therm and its significance are displayed. The Raw Δ column shows the average difference between thermophiles and mesophiles in raw percentage points; Δ gives the equivalent in standardised scores. The last columns report the significance of the observed difference and the test statistics that have been used.
aApproximate figures.
bSee Appendix for discussion.
n/a, not applicable.