You may also need to check that you have a 64-bit machine. If you wish to use more than 4GB, you need to get the blat source and compile it yourself.
You can see what system you have with uname -a You can see what kind of compiled blat executable you have with the file command: file /somepath/blat -Galt Ar 4/26/2010 1:36 PM, scríobh Hiram Clawson: > Good Afternoon John: > > Your login shell may have default memory limits for processes to > avoid run-away procedures from hogging all your memory. > > If you are in bash, you can raise your shell limits with the commands: > > ulimit -d 10240000 > ulimit -v 10240000 > > To set 10 Gb memory limits. If you are working in the C-shell (csh/tcsh) > use the limit command: > > limit datasize 10240m > limit vmemorysize 10240m > > The usual method to avoid this problem is to break up your alignment > problem into smaller parts. For example, align each of your query > sequences to a single chromosome in the target genome. You can > specify a single chromosome target sequence, for example: hg19.2bit:chr1 > to blat just to chr1 in hg19. > > --Hiram > > John Herbert wrote: >> Dear Mailing list, >> Please can someone help me fix a problem using a local version of BLAT? I >> want to align nucleotides to the human genome. >> >> I have this version: >> blat - Standalone BLAT v. 34 fast sequence search command line tool >> >> My workstation has a Core i 7 processor and 12Gb of ram. >> >> I indexed the genome doing: >> faToTwoBit hg18.fa hg18.2bit >> >> However, if I take a single sequence and align it with blat, this happens: >> >> j...@john-desktop:~/data/genome$ blat -out=blast9 hg18.2bit test >> test_seq_blas9.out >> Loaded 3107677273 letters in 49 sequences >> needHugeMen: Out of huge memory - request size 957189248 bytes >> >> If I am looking right, this amounts to 1Gb of memory but the machine has >> 12Gb. >> >> Is there a line of code in the source I can modify to get around this >> problem or am I missing something? >> >> Thanks for any advice. >> >> ps. I saw other similar posts but they have work arounds and as I have a lot >> of data to analyse, it would be great to sort it out so it works. >> >> John. > _______________________________________________ > Genome maillist - [email protected] > https://lists.soe.ucsc.edu/mailman/listinfo/genome _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
