Structure similarity whose scale does not drift with chain length

RMSD has no fixed meaning across lengths: 3 angstrom is excellent for a 300-residue protein and poor for a 40-residue one. The d0 normalisation fixes that, so a TM-score above 0.5 means the same fold at any size. The distance weighting also caps the damage a few badly placed loops can do.

Usage

Tmscore(coords1, coords2, l_ref = NULL)

Arguments

coords1, coords2

Aligned, already superposed residue coordinates.

l_ref

Reference length; number of aligned residues by default.

Value

List with estimate, d0, rmsd, L, L_ref.

Details

Formula: TM = (1/L_ref) sum_i 1/(1 + (d_i/d_0)^2), d_0 = 1.24 (L_ref - 15)^(1/3) - 1.8.

References

Zhang, Y. & Skolnick, J. (2004). Proteins 57:702-710, equations (2) and (3).

Examples

V <- c(1, 2, 3, 4, 5, 6, 7, 8)
Tmscore(V, V)
#> $estimate
#> [1] 1
#> 
#> $d0
#> [1] 0.5
#> 
#> $rmsd
#> [1] 0
#> 
#> $L
#> [1] 8
#> 
#> $L_ref
#> [1] 8
#> 
#> $method
#> [1] "TM-score structural similarity"
#> 
#> attr(,"class")
#> [1] "morie_rich_result" "list"