samtools invalid BAM binary header

[bob-loblaw]

Hi all,

I've already done some googling on this and I haven't been able to find an answer to this incredibly infuriating problem!

So I have 12 sam files, which I need to convert to bam, then sort, index etc. All 12 were processed in the same way, but samtools is having difficulty with 3 of them, specifically it gives this error when I try to sort

-bash-3.2$ samtools sort x.bam x.sorted
[bam_header_read] invalid BAM binary header (this is not a BAM file).
[bam_sort_core] truncated file. Continue anyway.
Segmentation fault

Samtools was used to generate the bam files so it's strange that it can't recognize a file that it's generated itself. I've tried remaking the sam file using samtools view, same problem. I've tried regenerating the sam file (bowtie2) same problem again.

Anyone got any advice? this has been driving me a bit mad for the past 2 days.

[dpryan]

Did you create the BAM file without a header? Presumably you did a "samtools view -Sbo x.bam x.sam" instead of "samtools view -hSbo x.bam x.sam" (or something like that). In that case, I would just reheader or recreate the BAM file. As a general rule, try "samtools view" on any file giving you errors like this.

[Richard Finney]

Code in Heng Li's samtools source file bam.c is this ...

       magic_len = bam_read(fp, buf, 4);
        if (magic_len != 4 || strncmp(buf, "BAM\001", 4) != 0) {
                fprintf(stderr, "[bam_header_read] invalid BAM binary header (this is not a BAM file).\n");
                return 0;
        }

This comes after checking for EOF to see if it is truncated. So file is not a valid BAM file. Since 'B','A','M',0x1 is not in the first 4 bytes, it must be compressed.

Try this

ls *.bam | awk '{print "echo "$1";od -c -N4 "$1}' | bash

to see if the bams all have same first few bytes. You should see this ...

0000000 037 213 \b 004

Some other checking is find out what size the bam files are.

Also run this little command to see what "file" program says :

file *.bam

You should something like this example ...
[xxxx@biowulf]$ file *.bam
98017.bam: gzip compressed data, extra field
98018.bam: gzip compressed data, extra field
98029.bam: gzip compressed data, extra field
98031.bam: gzip compressed data, extra field
....

[meetelbeet]

bam file with no header upon conversion from sam

Hi everyone
I have a similar problem. I ran bowtie2 and filtered the sam file to get alignments with only certain fragment lenghts, I did this with awk:

awk '($9>=74 || $9 <=-74) {print}' ctrl-new.sam > noovctrl.sam

Then I converted the resulting sam file to bam:

samtools view -bhS noovctrl.sam > ctrl-new.bam

But this file has no header. When I try to look at its 'head' with:

samtools view ctrl-new.bam | head

I get the error:

[bam_header_read] bgzf_check_EOF: Invalid argument
[bam_header_read] invalid BAM binary header (this is not a BAM file).
[main_samview] fail to read the header from "ctrl-new.bam"

I have tried to cat the header, reheader... and nothing. By the way, earlier I did some other unix manipulations of the same original sam file and was able to convert to bams with no problems, so I don't think the problem is in the alignment file.

I would appreciate any advice, thanks to all

[swbarnes2]

I think the problem is that your awk statement is resulting in a .sam with no header, and I don't think .samtools view generates one, it just keeps the one that should already be there.

Start with your initial sam with its intact header. Make the header its own file with

samtools view -H original.sam > header.txt

Then cat that to the top of your awk-generated .sam, and now it should get converted.