bwa index error

[seq-nerd]

Hi, I'm having trouble indexing my genome. Here is the command I am using:

bwa index -a bwtsw /data/Bioinfo/genomes/hg18.fa

Here is the error:

[bwa_index] Pack FASTA... [bns_fasta2bntseq] fail to open file '/data/Bioinfo/genomes/hg18.fa.pac'. Abort!
Aborted

I have all the various hg18.fa files in the same directory as well:

hg18.fa
hg18.pac
hg18.ann
hg18.amb
hg18.rpac
hg18.bwt
hg18.rbwt
hg18.sa
hg18.rsa

I don't know what I'm doing wrong. Any suggestions would be appreciated! Thanks!

[drio]

try /data/Bioinfo/genomes/hg18.fa instead of /data/Bioinfo/genomes/hg18.fa.pac

[Bruins]

the various files are missing .fa in their extension.

[seq-nerd]

I do not specify the .pac at the end. I don't know why the command is showing that. I specify only hg18.fa....any ideas?

[seq-nerd]

the .fa's are i just forgot to type them.....

[kmewis]

I'm getting an error trying to index the hg19 as well.

I have all the .fa files from the chromosomes, I concatenated them into hg19.fa. I then run:

bwa index hg19.fa

and I get:

shakti:~/Documents/Bioinformatics/hg19 kmewis$ bwa index hg19.fa
[bwa_index] Pack FASTA... 70.84 sec
[bwa_index] Reverse the packed sequence... 38.55 sec
[bwa_index] Construct BWT for the packed sequence...
bwa(1791) malloc: *** vm_allocate(size=3095678976) failed (error code=3)
bwa(1791) malloc: *** error: can't allocate region
bwa(1791) malloc: *** set a breakpoint in szone_error to debug
Bus error

What could this mean? I'm running this on a PowerPC Mac with 16gb of RAM.

Thanks!

[drio]

What does file /path/to/bwa tell you? Was bwa generated as a 64 bits binary? It seems bwa failed while allocating memory.

[kmewis]

What does file /path/to/bwa tell you? Was bwa generated as a 64 bits binary? It seems bwa failed while allocating memory.

I don't know what file path/to/bwa is supposed to tell me, it says:

/path/to/bwa: cannot open (/path/to/bwa)

My bwa executable is in ~/bin.

How can I tell if it was generated as a 64 bit binary instead of a 32 bit? I just used the makefile that came with bwa 0.5.8a...

[drio]

What do you get when you run?:

$ file ~/bin/bwa

[kmewis]

Here's the output:

shakti:~/Documents/Personal Stuff/Genome kmewis$ file ~/bin/bwa
/Users/kmewis/bin/bwa: Mach-O executable ppc

I'm using this on a PowerPC, so this is what I'd expect.

[drio]

It seems is 32 bits binary. The file command reports 64 bits in a 64bit system. Have you tried any other aligner or tool that requires to address big chunks of memory?

$ file ./bwa
./bwa: Mach-O 64-bit executable x86_64

Also, it would be useful is you post here the gcc directives you used while compiling (output of make).

[kmewis]

I haven't tried any other tools yet, though I've heard that bowtie is another good aligner to use.

I just tried the other alignment option, BWT, and it seems to be doing alright now.

bwa index -a bwtsw hg19.fa

I read that this is for single end reads, which I assume is what the hg19 sequence I downloaded is.

Thanks so much for the help though; just knowing that someone out there is willing to help gives me more motivation to try to get it working! I appreciate it. I'm sure I'll have additional questions once I've got this going, but these forums are very helpful.

[drio]

Be careful with bwtsw. That's the algorithm for longer reads. It may not be what you are
looking for.
I
think your system is not 64bits. You will have the same issues with other tools. I suggest you spend some time figuring that out.

[kmewis]

I don't think my system is 64 bits either; is a 64 bit system required?

It's been a couple weeks since I made bwa. It seems to have made properly, it gave me the bwa.exe file as an output. I made a new directory to try to make bwa again to give you the output, but it seems to not be working. Output is:

shakti:~/Documents/bwa-0.5.8a kmewis$ makefile
/Applications/muscle3.6_src/makefile: line 12: CFLAGS: command not found
/Applications/muscle3.6_src/makefile: line 13: LDLIBS: command not found
/Applications/muscle3.6_src/makefile: line 16: OBJ: command not found
/Applications/muscle3.6_src/makefile: line 17: EXE: command not found
RM: =: No such file or directory
RM: rm: No such file or directory
RM: -f: No such file or directory
CP: =: No such file or directory
/Applications/muscle3.6_src/makefile: line 22: GPP: command not found
/Applications/muscle3.6_src/makefile: line 1: GPP: command not found
/Applications/muscle3.6_src/makefile: line 1: CFLAGS: command not found
LD: can't open: = (No such file or directory, errno = 2)
LD: no object files loaded
/Applications/muscle3.6_src/makefile: line 1: GPP: command not found
/Applications/muscle3.6_src/makefile: line 1: CFLAGS: command not found
powerpc-apple-darwin8-gcc-4.0.1: =: No such file or directory
powerpc-apple-darwin8-gcc-4.0.1: warning: '-x c' after last input file has no effect
powerpc-apple-darwin8-gcc-4.0.1: no input files
/Applications/muscle3.6_src/makefile: line 26: all:: command not found
/Applications/muscle3.6_src/makefile: line 1: wildcard: command not found

If I understand you correctly, you're saying the indexing of the database is different depending if I'm going to be comparing longer (Sanger, 454) reads instead of short (Illumina) reads? What would be different between the two?

[drio]

1. You have to address more than 4G of memory, a 32bit system won't allow you to do that.

2. run make not makefile

3. Yes. The mapping strategies are different. There are two different papers explaining
both algorithms (sort/long reads).