Hi! If you use -out-blast9 (or any other blast non-psl format) then you will definitely not match the web-blat. This is because the blast output contains ONLY EXONS. The psl has nicely-chained alignments with exons and introns. You should use out=psl (the default) for that.
Of course, matching the webblat is not really a goal for a batch usage. webblat (hgBlat) is trying to show you virtually everying (up to limit of 200 alignments) so that an intelligent human can ponder the output. When doing a big batch of alignments, you just want to set the filtering as appropriate to your experiment and not look at every worthless low quality alignment. Using pslCDnaFilter works very well on your psls after the batch run and has many good options. -Galt Vinodh Srinivasasainagendra: > Many thanks for your recommendation. > > I used the parameter settings mentioned in anchored section #5 on the > page you recommended, but it spilts a ' needHugeMen' error message. I > have loads of mem (~20GB) available on this workstation where I am > performing blat. > > blat -stepSize=5 -repMatch=2253 -minScore=0 -minIdentity=0 -out=blast9 > hg18.2bit test_seq.txt test_seq_blas9.out > > needHugeMen: Out of huge memory - request size 2106194880 bytes > > > Is the only difference between web-blat and command-line blat is > 'repMatch' (command-line default is '1024' and web-based is '2253')? > > If the answer to the above question is YES, then what should I be doing > to handle this error message 'needHugeMen: Out of huge memory - request > size 2106194880 bytes'. > > Again, my execution line is ' blat -stepSize=5 -repMatch=2253 > -minScore=0 -minIdentity=0 -out=blast9 hg18.2bit test_seq.txt > test_seq_blas9.out' > > > vinodh > > -----Original Message----- > From: Hiram Clawson [mailto:[email protected]] > Sent: Thursday, February 11, 2010 4:08 PM > To: Vinodh Srinivasasainagendra > Cc: [email protected] > Subject: Re: [Genome] Blat related question, > > Good Afternoon Vinodh: > > Please note this discussion about operating blat: > http://genome.ucsc.edu/FAQ/FAQblat.html#blat5 > > --Hiram > > Vinodh Srinivasasainagendra wrote: >> Hello, >> >> I have a task in hand to perform sequence alignment using blat. >> Unfortunately, I have approximately 500+ individual sequences and the >> web-blat on ucsc's web-portal would not cater more than 25 at a time. >> For that reason, I went ahead and installed blat on my local server. > Ran >> the sequence alignment executable 'blat' against my Hg sequences and >> everything ran to completion using a command-line based run. But when > I >> compared my commad-line results with the web-based results from UCSC > for >> the same sequence, there seem to be some difference. >> >> I have not set any flags or configured my command-line version of the >> blat and it commands of execution looks like: >> >> blat -out=blast9 hg18.2bit test_seq.txt test_seq_blas9.out >> >> I tried the FAQ recommendation on the UCSC's blat web-space for the >> recommended flags to use to reproduce web-based blat results and get > the >> following error: >> >> blat -stepSize=5 -repMatch=2253 -minScore=0 -minIdentity=0 -out=blast9 >> hg18.2bit test_seq.txt test_seq_blas9.out >> >> needHugeMen: Out of huge memory - request size 2106194880 bytes >> >> I know for the fact I have loads of memory on my workstation. >> >> Is there a big algorithmic difference in the way the web-blat and >> commad-line default blat operate? >> >> For reproducible research purpose I would like for my command-line > blat >> to report the same results as the web-based blat (on ucsc's web > portal) >> >> Any recommendations, thoughts, comments will be appreciated. >> >> Many Thanks, >> >> Vinodh > > _______________________________________________ > Genome maillist - [email protected] > https://lists.soe.ucsc.edu/mailman/listinfo/genome _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
