NeedleHaystack: a program for the rapid recognition of local structures in large sets of atomic coordinates

A program that computes molecular superpositions is presented. A possible large molecule (target = haystack) is scanned for the occurrence of a given molecular motif (model = needle) within tolerances. The advance over other methods is that it can handle very large atom sets, e.g. targets of up to 100000 atoms, and does not necessarily require any chemical information and connectivity between atoms: chemical constraints to restrict the solution space are not used to relate non-obvious superpositions based on a geometric fit. The specialization to a needle-in-a-haystack problem allows a very fast algorithm. Trade-offs between run time and tolerance levels are possible. The program is fast enough to screen large databases in an acceptable time. An executable for Linux on Pentium-compatible machines is available free of charge at http://www.charite.de/bioinf/haystack . It is a command-line executable well suited for scripts and distributed computing. A Web interface for immediate testing is also available at this address.