“PISCES is a publicly accessible web server used for generating representative protein structure or sequence lists”

PISCES is a publicly accessible web server used for generating representative protein structure or sequence lists. Such lists are used in statistical analysis of protein sequences and protein structures and for testing protein structure prediction methods. PISCES provides an easy-to-use interface for generating these lists based on a highly optimized protocol for comparing proteins, using both sequence and structure comparison. PISCES is used intensively by researchers worldwide.

PISCES uses PSI-BLAST and CE structural alignments to establish evolutionary relationships between PDB chains. Local alignments with PSI-BLAST are much more accurate than similar servers using pairwise sequence alignment and/or global alignment algorithms. PISCES identifies evolutionary relationships with PSI-BLAST and uses CE to determine alignments for the most distantly related pairs, thus ensuring accurate and complete alignments at sequence identities from 15-100%. We also provide highly annotated sequence files for proteins of known structure, that include functional information and species for each protein chain in each entry in the Protein Data Bank.

Proteins come in many shapes and sizes. Like different species of animals and plants, some proteins are closely related to one another, and others are very distantly related to one another. For many purposes, a researcher wants to have a set of proteins that are all dissimilar to one another. This is accomplished by eliminating some sequences from a list of all protein sequences of known structure until all pairs in the remaining set are at least dissimilar to one another by some specified degree. The proteins in such a list can then be used for statistical analysis or testing methods for predicting protein structures using computer programs.

I was a postdoc in Dr. Roland Dunbrack's lab at Fox Chase Cancer Center when I implemented the PISCES server based on earlier work in the Dunbrack group. Dr. Dunbrack's research involves developing methods of protein structure prediction and thus depends on the outputs of our PISCES server.