Here we show partiFold's reliability and accuracy by comparing predictions made against known contacts in PDB structures. All types of predictions that partiFold can make are included, as well as all datasets used for any portion of the algorithm. When not indicated, statistical potentials were computed using the Wang & Wang reduced alphabet.

Comparisons against 8 non-homologous, amphipathic TMBs

Results using single standard set of short-sequence/small-barrel constraints:

Results using single standard set of long-sequence/small-barrel constraints:

Results using protein-specific, hand-refined constraints:

Dataset evaluation

A list of PDB IDs used for determining statistical potentials is available here.