A method for constructing thousands of compact protein conformations from fragments and then connecting these structures to form a network of physically plausible folding pathways

We sample the conformational space of small proteins and link these conformations into a graph of physically plausible folding pathways. Conformational sampling is performed using HMMSTR, a hidden Markov model for local sequence-structure correlations. Our method uses only the amino acid sequence to bias the conformational sampling; no knowledge of the native structure is used. This sampling strategy drastically reduces the size of the confor- mational space to be searched. We then build a probabilistic roadmap (PRM) graph and find paths that have the lowest energy climb. We find that favored folding pathways exist, corre- sponding to deep valleys in the energy landscape. In contrast to previous PRM methods that used knowledge of the native structure to sample conformational space, this is the first at- tempt to merge the previous successes in fragment assembly methods with PRM methods.