Table 3. Effect of sequence-structure mapping.

Domain†	Alignment(best peptide)*	Without alignment (best peptide)**
ABP1	0.39	0.36 (−3, II)
Amphyphisin	0.43	0.47 (−1, II)
Endophilin	0.53	0.54 (−1, II)
MYO5	0.36	0.31 (−3, I)
RVS167(I)	0.48 (0.25) ‡	0.34 (−3, I)‡
RVS167(II)	0.48	0.48 (0, II)
SHO1(I)	0.52	0.50 (−3, I)
SHO1(II)	0.26 (0.24) ‡	0.24 (0, II)‡
LSB3(I)	0.56 (0.44)‡	0.43 (−3, I)‡
LSB3(II)	0.71	0.69 (0, II)
YSC84(I)	0.40	0.38 (−3, I)
YSC84(II)	0.57	0.57 (0, II)

The Pearson's correlation coefficients between the predicted binding energies and SPOT data are shown.

^†When sequences are separated into Class I and Class II, the class is marked in parentheses. Abp1, Amphyphisin, Endophilin, and Myo5 do not have the canonical SH3 motifs.

*Pearson's correlation coefficient for the best peptide template when alignments are adjusted.

**Pearson's correlation coefficient for the best template peptide when the alignment is fixed to that of canonical motif PxxP. The offset and class of peptide templates are indicated in parentheses.

^‡Cases when the class of the best peptide template is inconsistent with the class of sequence motifs. The best peptide belonging to the sequence motif is indicated in parentheses in the second column. The correlation of fixed alignment for that peptide is shown in the third column.