If a user submits a one-sequence query where the identifier and sequence are not separated by a newline, Seqserv crashes.

E.g., sometimes users incorrectly store what should be the following query fasta:

>identifier123
AGAGCTAGCTAGCTACGATCGATCGATGCAAGTAGTACtAGCTCGA

They might store it (and paste it into seqserv's query like so:

>identifier123 AGAGCTAGCTAGCTACGATCGATCGATGCAAGTAGTACtAGCTCGA

This overcomes our internal checks - it's identified as nucleotide in Seqserv's javascript in-browser testing, and starting with '>', it's identified as FASTA. But it's not correct FASTA. BLAST sees this as being an identifier and a long description, and no sequence.

Blast will provide this type of error:

We should elegantly catch that and return that to the user, rather than just failing (we currently fail with "there is a problem - likely with BLAST binaries"

I don't think this occurs that frequently. Two options:

• extra FASTA validation on the client side (would want to be lightweight)
• better error catching when BLAST fails, and reporting BLAST's server-side error to the client.

My intuition is that the 2nd is better.