Different results between local and online BLAST (RefSeq database)

Question

[philipp]

Hello,

I BLASTed a sequence online and on my local machine against the RefSeq_genomic database. Online I get a query coverage of 17% and 74% identity. My local BLAST (same parameters) finds a hit (gi 330443688) with a query coverage of 98% and 72% identity. The second odd thing is that if a align my query sequence to gi 330443688 the online BLAST gives the same numbers.

My question is why does the online BLAST not find the "better" hit? And which result is more reliable?

I appreciate your help!

Best Philipp

Query sequence: ATGAGCGAATTACGTGCATTCAGCGCGCCAGGTAAGGCACTGCTGGCCGGTGGCTACCTGGTGTTAGACACCAAGTACGAGGCGTTCGTCGTCGGCTTATCTGCCCGTATGCATGCAGTTGCCCACCCGTATGGTAGCCTGCAGGGCTCTGACAAGTTCGAAGTGCGTGTGAAGAGCAAGCAGTTCAAGGACGGCGAGTGGCTGTACCACATTAGCCCAAAGAGCGGCTTCATCCCGGTTAGCATTGGTGGCAGCAAGAACCCATTTATCGAGAAGGTCATTGCCAACGTCTTCAGCTACTTCAAGCCGAATATGGACGATTACTGCAACCGCAACCTGTTCGTCATCGACATTTTCAGCGACGACGCGTACCACAGCCAAGAGGACTCTGTTACGGAGCATCGTGGTAACCGCCGCCTGAGCTTCCACAGCCATCGCATTGAGGAGGTGCCGAAGACGGGTCTGGGTTCTAGCGCCGGTTTAGTTACCGTCTTAACGACGGCGTTAGCGAGCTTCTTCGTGAGCGACCTGGAGAACAACGTGGACAAGTACCGCGAAGTGATTCATAACCTGGCGCAGGTGGCACATTGTCAGGCCCAAGGTAAGATTGGCTCTGGTTTTGATGTGGCAGCGGCCGCCTATGGCTCTATCCGCTATCGCCGCTTTCCGCCGGCCCTGATCAGCAATCTGCCGGACATCGGCTCTGCGACGTATGGTAGCAAACTGGCGCATCTGGTGGACGAAGAAGACTGGAACATCACCATTAAGTCTAATCACCTGCCGAGCGGCTTAACGTTATGGATGGGCGATATCAAGAACGGCAGCGAAACGGTTAAGCTGGTGCAGAAAGTGAAAAACTGGTACGACAGCCACATGCCGGAAAGCCTGAAGATTTACACGGAGCTGGACCACGCCAATAGCCGTTTCATGGATGGTCTGAGCAAGCTGGACCGCCTGCACGAAACCCACGACGACTACAGCGACCAAATCTTCGAGAGCCTGGAGCGCAATGACTGCACCTGCCAGAAGTACCCGGAGATCACGGAGGTCCGCGATGCCGTGGCAACGATTCGCCGTAGCTTCCGCAAAATTACGAAGGAGAGCGGCGCGGATATCGAACCACCGGTCCAGACGTCTCTGCTGGACGACTGTCAAACCTTAAAGGGCGTGTTAACGTGCCTGATTCCGGGCGCGGGTGGTTACGACGCCATTGCCGTCATCACGAAACAGGACGTCGATCTGCGCGCACAAACGGCCAACGACAAACGTTTCAGCAAAGTCCAATGGCTGGATGTTACGCAGGCCGACTGGGGTGTTCGCAAGGAGAAGGACCCGGAAACGTATCTGGATAAGTGA

[Raony Guimarães]

Are you sure you are using the same version of BLAST ?

[philipp]

On my local machine I run BLAST 2.4.0. I'm not sure which version is used on the NCBI website but I assume they use the new 2.6.0 update.

Do you think that explains the difference? I would argue that maybe the new code is more efficient and / or more stable but should give the same results.

I set up an AWS server and keep you updated which results I get there.

Best, Philipp