BLAT standalone version - why is it so slow?

Question

[kct913]

I am using BLAT for one of my class projects and have downloaded the standalone version's executables from http://hgdownload.soe.ucsc.edu/admin/exe/linux.x86_64/ and moved it to my class server. I ran BLAT using the following command:

~/blat_standalone/blat/blat ../GRCh38_latest_genomic.fna test_4.fa blat_out4.psl

As indicated above, I am running a test sequence against hg38. The test sequence is the mRNA sequence of POLR2A. When I ran the sequence on the web version of BLAT on UCSC it only took less than a second. However, when I tried to run the same sequence using my own standalone version, it took about 40 seconds. The standard output, while the program was running, is as follows:

  $ ~/blat_standalone/blat/blat ../GRCh38_latest_genomic.fna test_4.fa blat_out4.psl
  Loaded 3272089205 letters in 639 sequences
  Searched 6749 bases in 1 sequences

Question: 1) Why is my program so much slower than the web version and how do I make it faster? I suspect it may have something to do with the fact that I am not using the -ooc option and an ooc file. My understanding is that I need to create the file, but I don't know how to make one. 2) can someone please tell me how to create a 11.ooc file from my database? 3) It took around 14 seconds for the program to load all of the sequences from the database (all 639 of them). Will using the -ooc option reduce this time? If not, will deleting all but the 25 primary assemblies significantly reduce this time?

Answer 1

[Alex Reynolds]

I have a post here about setting up BLAT:

http://www.bitsumma.com/2017/09/03/how-to-install-and-set-up-a-local-ucsc-blat-environment/

The "Optimizing Queries" section shows how to generate an OOC file. Using it will indeed speed up queries.

This article also discusses parallelizing queries, which may or may not be relevant to your use case, but is also an approach to improve performance.