The BACH-server allows to use the BACHSCORE tool with the aim of discriminating the best model of a protein (the native state or a structure very similar to it) among a large set of alternative conformations.

The tool uses the BACH++ (Bayesian Analysis Conformation Hunt) knowledge-based potential, an all atom energy score based on 1091 parameters derived by a statistical analysis of a small set of experimentally observed proteins.

The original BACH scoring function was introduced in [1], whereas the BACH++ upgraded scoring function used in BACHSCORE is presented in [2].

The BACH-server provides direct access to the BACH++ energy function. Both single models and sets of models (submitted as a single zip file or collectively selected from a dropdown list) can be analyzed. Input model files need to be in pdb (Protein Data Bank) format.

An energy score is associated to all models in the set and then used to rank them: conformations with lowest energies are predicted to be more similar to the native state.

IMPORTANT WARNING: in the scoring function there are no energy terms taking into account steric clashes or unrealistic values of torsional angles, so that unrealistic structures could be assigned a low energy score.


[1] P. Cossio, D. Granata, A. Laio, F. Seno and, A. Trovato
"A simple and efficient statistical potential for scoring ensembles of protein structures"
Scientific Reports 2, Art. N. 050901 (2012)

[2] E. Sarti, S. Zamuner, P. Cossio, A. Laio, F. Seno, and A. Trovato
"BACHSCORE. A tool for evaluating efficiently and reliably the quality of large sets of protein structures"
accepted for publication in Computer Physics Communications

Source Code

The BACHSCORE source code, including scoring function parameter files, will be distributed by the Computer Physics Communications International Physics Program Library (Catalogue Identifier: AEQD_v1_0)