communications to NCBI, it's clear that there is definite interest in
having a NCBI FASTA mailserver. We had initially decided to forgo this
because of the demands on resources, and because, as mentioned by M.
Cherry, we had been making BLAST available to a number of sites on an
experimental basis and had not received explicit feedback regarding the
need for additional access to FASTA. Bill Pearson has generously
agreed to modify FASTA so that it can read the file format we use
centrally for sequence searching (this same format will be available on
CDROM's updated every 2 months starting in February). Thanks to Bill's
work, the single format will avoid the need for another copy of all the
sequence data on our BLAST server.
When this software is available and tested we'll put the FASTA server
up for public access and announce the availability via the bulletin
boards and in the header messages of our e-mail servers. The priority
of these searches will probably be lower than BLAST searches due to
their greater requirements for CPU cycles.
Supporting additional computational tools makes real demands on our
resources and have to be balanced by significant gains in
functionality. For example, I'm not convinced that FASTN searches are
really worth adding -- although they result in alignments with gaps,
they are unlikely to yield matches that would not be found with
BLASTN. In order to detect significant nucleic acid similarities in
such large databases requires fairly strong similarity (coding or
non-coding) and these non-gapped regions would be found with BLASTN.
This, of course, is arguable and your input is appreciated.
We will be doing our best to continue to upgrade the sequence searching
facilities at NCBI, incorporating new computational approaches developed
at NCBI and from researchers around the world. Suggestions are welcome!