SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
using bwa to map illumina paired end reads mikeworth Bioinformatics 6 08-13-2013 06:02 PM
PubMed: Fragmentation of contaminant and endogenous DNA in ancient samples determined Newsbot! Literature Watch 0 09-11-2011 08:40 AM
Illumina mapping with bwa bpetersen Bioinformatics 5 03-14-2011 11:04 PM
18bp DNA contaminant in ChIP input sample? sinapius Sample Prep / Library Generation 0 02-09-2011 11:39 AM
Need help defining BWA parameters for human screening jmartin Bioinformatics 4 04-02-2010 03:52 PM

Reply
 
Thread Tools
Old 12-09-2009, 06:32 PM   #1
jmartin
Member
 
Location: St. Louis

Join Date: Dec 2009
Posts: 72
Default BWA and screening illumina reads for contaminant

I'm trying to hash out a set of parameters for bwa that I can use to screen illumina 75mer & 100mer reads for contaminant. I'm expecting a large volume of environmental data, and I want to clean the incoming data by filtering out certain organisms.

I'm worried that the default parameters for bwa may prove too stringent. But I'm not sure which parameters to relax without introducing too many false positives into my results.

I have a set of data spiked with contaminant to use for testing, but I don't really know where to start. The bwa man page says setting an INT value for '-n' sets the "maximum edit distance", and that seems to set the max number of mismatches allowed. And I can watch results shift as I vary that number, and in most cases it seems to make sense. But that seems at odds with the arguments '-k' & '-l' with are supposed to set the seedlength (which is by default 'inf' which I guess implies the full length of my query), and the max "edit distance" (mismatches?) in the seed (default is 2). But I can vary the '-n' value and allow more than 2 mismatches even though I'm using the default settings for '-k' & '-l' (which should kill the read if there are more than 2 mismatches?)

Is there any more detailed discussion of parameter suggestions for real world problems? I'm pretty old school, and BLAST had that wonderful oreilly blast book that had a number of examples that gave you a starting point for different tasks, then I could tweak values from there. But I feel kinda lost with BWA. Can anyone suggest another resource beyond the man page & the bwt paper that might help me out?
jmartin is offline   Reply With Quote
Old 12-09-2009, 07:40 PM   #2
lh3
Senior Member
 
Location: Boston

Join Date: Feb 2008
Posts: 693
Default

I am not familiar with metagenomics analysis, but probably you would like to try "bwa bwasw" if you think "bwa aln" is too stringent. Bwasw simply gives you high-scoring local hits (kind of like blast/ssaha2) rather than require the full read to be aligned (bwa aln requires this). I do not know if this meets your goal. BTW, be aware that the total length of your database sequence must smaller than 4GB. If you want to align against the nt database, split it first.
lh3 is offline   Reply With Quote
Old 04-22-2010, 03:28 PM   #3
xchen5
Junior Member
 
Location: georgia

Join Date: Mar 2010
Posts: 3
Default

I have the same question:
I guess in BWA input, "-l 25 -k 2" means that if there are more than 2 mismatches (or differences from the reference sequence) in the first 25 base (called seed?) of a read, the read will be considered as failed to map.

Dr. Liheng, is this right?

Last edited by nilshomer; 04-22-2010 at 06:52 PM.
xchen5 is offline   Reply With Quote
Old 04-22-2010, 03:55 PM   #4
Auction
Member
 
Location: california

Join Date: Jul 2009
Posts: 24
Default

Quote:
Originally Posted by jmartin View Post
I'm trying to hash out a set of parameters for bwa that I can use to screen illumina 75mer & 100mer reads for contaminant. I'm expecting a large volume of environmental data, and I want to clean the incoming data by filtering out certain organisms.

I'm worried that the default parameters for bwa may prove too stringent. But I'm not sure which parameters to relax without introducing too many false positives into my results.

I have a set of data spiked with contaminant to use for testing, but I don't really know where to start. The bwa man page says setting an INT value for '-n' sets the "maximum edit distance", and that seems to set the max number of mismatches allowed. And I can watch results shift as I vary that number, and in most cases it seems to make sense. But that seems at odds with the arguments '-k' & '-l' with are supposed to set the seedlength (which is by default 'inf' which I guess implies the full length of my query), and the max "edit distance" (mismatches?) in the seed (default is 2). But I can vary the '-n' value and allow more than 2 mismatches even though I'm using the default settings for '-k' & '-l' (which should kill the read if there are more than 2 mismatches?)

Is there any more detailed discussion of parameter suggestions for real world problems? I'm pretty old school, and BLAST had that wonderful oreilly blast book that had a number of examples that gave you a starting point for different tasks, then I could tweak values from there. But I feel kinda lost with BWA. Can anyone suggest another resource beyond the man page & the bwt paper that might help me out?
In my recent experiments, I tried both BWA SW (default option) and BLASTN (1e-5) to filter out 454 reads from certain organism. It seems the reported number is very close.
Auction is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 09:30 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO