View Single Post
Old 02-26-2015, 11:54 AM   #12
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,695
Default

Hi Vicente,

I just tested BBSplit and it worked fine for me:
Code:
bbsplit.sh -Xmx1g ref=Clostridium_perfringensATCC_13124.fasta,Desulfotomaculum_gibsoniae_DSM_7213.fasta,Hirschia_baltica_ATCC_49814.fasta,Nocardiopsis_dassonvillei_DSM_43111.fasta

randomreads.sh -Xmx1g reads=1000 out=reads.fq

bbsplit.sh -Xmx1g in=reads.fq basename=o_%.sam
produced 4 output files:
Code:
[email protected]:/global/scratch2/sd/bushnell/splittest$ ls -l
total 38144
-rw-rw-r-- 1 bushnell bushnell  3311027 Feb 26 10:24 Clostridium_perfringensATCC_13124.fasta
-rw-rw-r-- 1 bushnell bushnell  4952723 Feb 26 10:24 Desulfotomaculum_gibsoniae_DSM_7213.fasta
-rw-rw-r-- 1 bushnell bushnell  3599251 Feb 26 10:23 Hirschia_baltica_ATCC_49814.fasta
-rw-rw-r-- 1 bushnell bushnell  6652569 Feb 26 10:23 Nocardiopsis_dassonvillei_DSM_43111.fasta
-rw-rw-r-- 1 bushnell bushnell    81824 Feb 26 11:43 o_Clostridium_perfringensATCC_13124.sam
-rw-rw-r-- 1 bushnell bushnell   125594 Feb 26 11:43 o_Desulfotomaculum_gibsoniae_DSM_7213.sam
-rw-rw-r-- 1 bushnell bushnell    81422 Feb 26 11:43 o_Hirschia_baltica_ATCC_49814.sam
-rw-rw-r-- 1 bushnell bushnell   181854 Feb 26 11:43 o_Nocardiopsis_dassonvillei_DSM_43111.sam
-rw-rw-r-- 1 bushnell bushnell   352650 Feb 26 11:42 reads.fq
drwxrwsr-x 4 bushnell bushnell      512 Feb 26 10:24 ref
When you say "without success", what exactly happens? And by the way, to catch unmapped reads, you need to set "outu1" and "outu2", not "out1" and "out2".

As for your other question - you can do quality-trimming before or after, but error-correction should be done after. Generally, I would do quality-trimming before if you use BBSplit's default settings, which require high identity, or after if you reduce the identity threshold to, say, minid=0.75 (which is closer to BBMap's default). But error-correction is best done after, on each file individually, so that homologous areas shared between the genomes don't affect each other.
Brian Bushnell is offline   Reply With Quote