HMMER searches
Sean Eddy's popular HMMER algorithm has been accelerated 30 times on the Bioinformatics Cell. Profile hidden Markov models (profile HMMs) are used to do sensitive database searching using statistical descriptions of a sequence family's consensus. HMMER is an implementation of profile HMM software for protein sequence analysis.View our HMMER fact sheet
Smith-Waterman searches - when you need ALL the answers
Finding homologous DNA, RNA, or protein sequences in a database can be done in many ways. Smith-Waterman based searches is the only method that identifies all true hits, but the algorithm is very slow when working on large datasets and most scientists therefore use the much faster BLAST search algorithm.But the speed of BLAST comes at the expense of the quality. The sensitivity of BLAST is in fact so low, that there is a significant risk of missing important sequences of interest. Compared to Smith-Waterman based searches, up to 50% of the search hits are thus not found using BLAST.
With the Cell, you can speed up a Smith Waterman search previously taking 2 hours to around 1 minute, and the Cell thus removes the need for using BLAST - at least in situations where you search through data where you not only need some of the answers but all of the answers.
The Cell includes the fastest Smith-Waterman implementation ever made on standard hardware - nucleotide searches are accelerated up to 130 times and protein searches are accelerated up to 75 times on most modern computers.
ClustalW alignments
ClustalW is one of the most used methods for doing multiple sequence alignments. When aligning many sequences and when aligning long sequences, the speed of the algorithm is however not impressing.As ClustalW is so widely used, we have implemented a SIMD-version of it in the Cell, resulting in an acceleration of 10 times on most modern computers.
