SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Bioinformatics (http://seqanswers.com/forums/forumdisplay.php?f=18)
-   -   About script to combine steps of BWA (http://seqanswers.com/forums/showthread.php?t=12224)

CNVboy 06-21-2011 03:13 PM

About script to combine steps of BWA
 
For BWA, first do alignment (generate .sai), then sampe the paired-ends (generate .sam), then convert .sam to .bam. Now I'm going to write script to combine all these steps. ie, input is fastq, then output is .bam file.

Now a big question is, I need to guarantee, say, the first step (generating .sai) is COMPLETE (instead of being aborted) before going to the second step; and the second step (generating .sam) before the third. I there any way to check that for BWA? Or what's the sign for the completeness of each step?

btw, for combining of steps of BWA, I just need to put those three commands together? Or any other trick?

thanks!

dp05yk 06-21-2011 08:22 PM

Just write a shell script that calls your three commands in order... unless you are executing BWA on a server with a scheduler, in which case you can just submit all three commands independently with wait conditions.

Jon_Keats 06-21-2011 09:24 PM

If you are looking for one to cut and paste there is one on my introduction thread that will take samples from fastq files straight to sorted bams with dupsmarked. It doesn't do a check for errors after each step, but others could likely suggest a simple method to capture if the exit status of the previous step. Generally, I've not had many issue with errors except when I messed up...

gprakhar 06-21-2011 11:17 PM

Hello CNVboy,

If you are running your combined script on a Scheduler then, only problem with giving all the commands together is that you will get to know about the errors only at the end, that too only "IF" you are not keeping an eye on the log file.
(use "$tail -f <filename>" for that, keep looking at it every 10-15 mins)
I use SGE and just tail -f the err and out files

Same way for running the script directly, only difference is your output is now the terminal and not a file.

Hope this helps,
prakhar

CNVboy 06-22-2011 05:13 AM

THanks everyone.

I think the core question is: how can we guarantee the quality of the generated .sai, .sam file? How do you know the programming at this step is complete? Just look at the size of files; or ,say judge the quality of .sai file by checking if it can produce .sam file?

For example, some of my .sai file looks good size-wise, but somehow cannot produce .sam file with correct size (.sam file is only 8.0kb); then I just reproduce .sai file, and it can generate .sam file with correct size.

Also, seems BWA returns exit codes; can we use this as the sign of completeness of each step? And try to make it like, command 2 will execute only if command 1 runs successfully

dp05yk 06-22-2011 06:16 AM

In my experience BWA won't really ever error out unless there's a problem with your data or your command (or lack of memory). As long as your data/commands are correct you can just use a shell script - command n in a shell script will wait until command (n - 1) is complete before running so you're guaranteed to start in the correct order.

As far as you yourself knowing if the .sai stage is complete, just check your stderr output - if it says "n sequences have been processed", where n = the number of sequences in your FASTQ file, then it's complete. But the shell will check this for you if it's running a script.


All times are GMT -8. The time now is 10:40 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.