Searching with a short domain

Results of searching SwissProt using SSEARCH (Smith-Waterman) with gap penalties of -14, -1 using a 76 residue putative calcium-binding domain from "HSU06863_1 follistatin-related protein precursor". Now the calmodulin type EF-hand domains are very well represented, although in this case only one has statistically significant similarity.

NCS1_RAT NEURONAL CALCIUM SENSOR 1 (NCS-1).         107 163.3  0.012
TPCS_RANES TROPONIN C, SKELETAL MUSCLE.              96 147.6  0.089
VIS3_RAT VISININ-LIKE PROTEIN 3 (VILIP-3) (NEURAL    92 140.0   0.24
CABI_BOVIN VITAMIN D-DEPENDENT CALCIUM-BINDING PRO   86 138.5   0.29
YLJ5_CAEEL HYPOTHETICAL CALCIUM-BINDING PROTEIN C5   89 136.1   0.39
LPSA_LYTPI CALCIUM-BINDING PROTEIN LPS1-ALPHA.       92 135.7   0.41
BDR1_HUMAN CALCIUM-BINDING PROTEIN BDR-1.            89 135.4   0.43
SP2A_STRPU SPEC 2A PROTEIN.                          86 132.9   0.59
SP2C_STRPU SPEC 2C PROTEIN.                          86 132.8   0.59
TPCS_RABIT TROPONIN C, SKELETAL MUSCLE.              86 132.4   0.63
CALG_PIG CALGIZZARIN (S100C PROTEIN).                83 131.8   0.67
CALG_RABIT CALGIZZARIN (S100C PROTEIN).              83 131.6   0.70
S10P_HUMAN S-100P PROTEIN.                           82 130.6   0.79
TPCS_HUMAN TROPONIN C, SKELETAL MUSCLE.              84 129.3   0.93
PSOR_HUMAN PSORIASIN.                                81 128.6   1.00

This is clearly an example of a sequence relationship that is best appreciated retrospectively, with the data from the tertiary structure. Without that evidence, one could not conclude that these two types of Ca++-binding proteins were homologous.