We have several local services to help you discover significant similarities between your query sequence and other sequences stored in the databases. BLAST & BLAST2.0 This method compares your sequence to all those present in the protein or nucleotide sequence databases, using a fast algorithm. Be aware, however, that BLAST has two limitations: (1) it pre-filters the databases by using a word matching procedure: this means that it can miss significant similarities if they do not contain a small region of high similarity. (2) although the WU BLAST scoring method produce gapped alignments, it is often advisable to extend the alignments by using more flexible tools, such as lalign (W. Pearson) or the GCG gap or bestfit programs. Please read the BLAST documentation if you are not sure of what it does. We have installed both the NCBI an implementation of BLAST2 on a 4-processor Pentium Pro machine and the Washington University (Warren Gish) implementation of BLAST on a 2-processor Pentium Pro machine at ISREC. The WU version produces gapped alignments, and is thought to be slightly more sensitive than NCBI BLAST. At ISREC, the BLAST2.0 version of the NCBI runs on a 4-processor Pentium Pro machine, and is therefore significantly faster than WU-BLAST. Hardware-accelerated Smith-Waterman (GeneMatcher) This is an implementation of the original Smith-Waterman dynamic programming algorithm. Because this algorithm is too slow to run on conventional computers, it uses a specialized processor, the Fast Data Finder (FDF), for its operation. The searches, especially for the frame searches, still take several minutes. Therefore, the FDF will preferentially send you the results by e-mail. PROSITE pattern search This searches your protein sequence for the occurrence of patterns ("consensus sequences") defined in the PROSITE database. Profile scan This searches your protein sequence for matches with "profiles" that represent essentially all of the information that can be derived from the alignment of a set of sequences. The profile method is, in our opinion, the most sensitive way to detect the presence of shared domains in protein sequences. You are advised to interpret with caution the results obtained with profile databases other than PROSITE. ny highly conserved protein domains, it may not detect domains that are less well conserved. SAPS (Statistical Analysis of Protein Sequences). This program, written by the group of Samuel Karlin, analyses proteins for statistically significant features like charge-clusters, repeats, hydrophobic regions, compositional domains etc. One of its options is to generate self-explanatory output. TMPRED (Prediction of transmembrane regions). This program tries to find putative transmembrane domains in proteins and also speculates on the possible orientation of these segments. For its scoring, it uses a combination of multiple weight-matrices that have been extracted from a statistical analysis of TMbase, a collection of all annotated transmembrane proteins present in SwissProt. BOXSHADE 3.21 (pretty printing of multiple alignments). This program generates publication-quality (or editable) graphics from multiple-alignment files. Many different input- and output-formats are supported. Conserved and/or similar residues are emphasized by various degrees of shading. It is also possible to specify a similarity-threshold that has to be reached for shading a residue. PatternFind (Search a protein database with a pattern). This PERL-Program takes a user-defined pattern (PROSITE-format or regular expression) and searches a protein database. It offers several useful output options. COILS (Prediction of coiled coil regions). This program predicts (2 stranded) coiled coil regions in proteins by the Lupas-algorithm. Some of these programs, as well as TMbase and some PC-based programs are available from our ftp-server.