Automatic classification of protein structures using low-dimensional structure space mappings

BMC Bioinformatics

Table 1 Alignment scoring methods.

Method	Raw score	Probability score	Accounts for	Number of free parameters
CE	Sum of non-linear AFP scores	Fits scores to Gaussian distribution	Statistical significance	1 (μ = 0)
Dali	Non-linear matrix similarity function	Division by polynomial fit to length dependence of alignment score [21]	Length	4
FATCAT	Sum of non-linear AFP scores	Fits scores to extreme-value distribution [22]	Statistical significance	2 (ζ = 0)
MATT	Sum of non-linear AFP scores	Non-linear function of RMSD, chain and alignment lengths [11]	Statistical significance, length	3

Summary of key factors associated with the scoring strategies of CE, DALI, FATCAT and MATT. The rightmost column gives the number of free parameters in the probability scoring function or assumed probability distribution, not including raw score or alignment length. Parameters are considered "free" if they have to be estimated empirically or evaluated using a large set of structure alignment scores as a reference.

ISSN: 1471-2105