QuickBLASTP, an accelerated version of BLASTP, adds a new pre-processing step to the non-redundant (nr) protein database. In a matter of seconds, QuickBLASTP will find approximately 97% of the database sequences with 70% or more identity to your query and around 98% of the database sequence with 80% or more identity to your query.
Currently, QuickBLASTP will only accept searches with a total query length less than 10,000 residues. You may only search the nr database with QuickBLASTP.
Quite interesting. Does exits the possibility of downloading the corresponding executables so we can try it locally? Running times for regular blastp over nr database are becoming prohibitive, so we’re testing alternatives as MMseqs2, DIAMOND or SWORD that cut down the search time between 5x-10x, while having the database in external storage. In optimal conditions, with the indexed database of MMseqs2 in cache, we achieve an speed up > 50x, setting the sensitivity parameter to 6.
@abiadak if you set MMseqs2 to -s 3 you should detect most of the hits at at 80%-70% sequence identity and the speedup will be huge.
Any chance of a more technical description? Blog or, better yet, a paper or something? While QuickBLASTP is indeed quick and seems ideal for a quick “What this sequence even is” kind of search, it may be important to know what are the exact optimizations behind it.