Indexing reference genome for GATK

Question

[lcc1844]

I have created .bam files using BWA and Samtools having downloaded a reference genome hg19 from UCSC and indexing it with

bwa index -a bwtsw hg19.fa

This worked fine for me, until now when I am trying to call variants in GATK and see that I need a reference in karyotypic order. I have downloaded the hg19.fa.fai and .dict files from the GATK FTP site and have also downloaded their full reference genome which is in chromosome order. I read that it wad easier to do this than to try and order the indexed reference I already have.

Only I am now trying to index the new reference in the same way and get the message:

[bwa_index] fail to open file '‐a bwtsw' : No such file or directory

Any ideas what I should be doing?!

Thanks

[Devon Ryan]

What version of bwa and what is the exact command that causes the error message?

[lcc1844]

Hi it is bwa-0.7.12 and the command was

bwa index -a bwtsw hg19.fa

[Devon Ryan]

Are you sure you didn't type:

bwa index "-a bwtsw" hg19.fa

or something like that? This isn't an error that I can reproduce without using an incorrect command.

[lcc1844]

100% sure, I used the same command I have used in the past so I don't know why it wont work. In the BWA directory I see bwtsw files with various suffixes. At the moment I am running index alone (without bwtsw) but I am not familiar with how this works for the human genome.

[lcc1844]

Oh ok brilliant, thank you. The index command alone is processing nicely so I wont question it any more!

Answer 1

[matted]

It looks like you may have copy-pasted the command from somewhere with a strange character encoding, or something like that. Your dash character isn't the same as the "regular" one:

>>> ord(u"‐")  # I copied yours here
8208
>>> ord(u"-")   # this is the one on my keyboard
45

If I copy your bwa command, I get:

bwa index ‐a bwtsw test.fa
[bwa_index] fail to open file '‐a' : No such file or directory

As Devon suggested, I need to group bwtsw to get the exact error you did. Maybe your space character is strange and effectively did that? That's a guess, though.

bwa index "‐a bwtsw" test.fa
[bwa_index] fail to open file '‐a bwtsw' : No such file or directory

In any case, you no longer need to specify -a with bwa, as it will auto-detect the best algorithm (the default is auto). Therefore, what you're running now should work fine regardless.

Edit: It looks like overall it's a Unicode problem. Your hyphen is a Unicode hyphen (8208 is 0x2010, which is a regular hyphen). It's not the same as an ASCII hyphen (ASCII 45, which is 0x002d). I imagine the -a bwtsw problem is because it used a Unicode space instead of the plain ASCII one. I don't know much about character encoding (and am glad!) and how desktop systems manage it when copying text between applications, but it looks like that caused these issues.

[Devon Ryan]

Wow, good catch on the dash character.

[Ram]

This is the reason I always use TextWrangler, check encoding and try again if I'm copy pasting commands.

Answer 2

[Ray Lee]

Change the order

bwa index hg.fa -a bwtsw

This works for me.