blat vs. blast

[dina]

Hello,
As I was advised here I tried blat instead of lst and I was really surprised. For example I checked a 1500 bp contig, and while blast found only very partial hit (about 300 bp), blat found a 99% hit to the whole contig. So.. shouldn't I trust blast? how this can be explained?
Thank you!!!!

[nickloman]

BLAST filter on?

[Zigster]

BLAT is much more "splicey" than BLAST

we'd have to see your output to make a call

[nickloman]

BLAT is much more "splicey" than BLAST

Yeah - I'd expect the results to be the other way round from that described typically (i.e. lots of separate alignments from BLAT, one big one in BLAST)

[lh3]

BLAST has this X-dropoff. If the alignment has a long gap or a long masked region (if you are not using -F F), blast will split it into two alignments. Actually if you apply -F F and chain the blast result, I guess you should get better alignment than blat.

[dina]

blast vs. blat-continued

I tried the blast with and without the filter and the result was the same; for a 1500 bp contig blast finds only the last 300 bp (this is the only hit) while blat finds the whole contig,,,

[nickloman]

Sounds a bit strange particularly if there is 99% nucleotide identity as you say.

Why not post up the BLAST and BLAT reports?

[dina]

blast

I now noticed that if I do blsat through the viewmapper of the ncbi I receive the same result as in blat ( the whole contig from existent sequencing data), but if I blast via the regular blast I receive only the annotated region which is only the end of my contig. Now I'm tatally confused.So, if I blast from

http://www.ncbi.nlm.nih.gov/genome/seq/BlastGen/BlastGen.cgi?taxid=9606

or from

http://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastHome

the results are different...

[dina]

refseq genomic vs.nr/nt

ok, I understood why it happens. It depends if the database is refseq_genomic or nr/nt. when I use nr/nt only a small region of the contig is found. When I use refseq genomic, the whole contig receives a hit, as in blat.
What is more correct to do then? I am interested to find an integrated virus. So, I have the data from the sequencing a genome (the reads and the contigs). Should I blast them with refseq genomic or nr/nt?