Samtools view -c Readcount Issue

[Lalit]

I'm using samtools to determine the total number of reads in a uniquely large bam file as shown below. I'm having a hard time interpreting the result. Anyone else receive a negative read count from samtools view -c before?

$ samtools view -c /run/media/data/Enclosure3/mergedcohort.bam
-1203401967

[sdriscoll]

Must be an overflow of whatever type they are using in the code. In other words the actual count is larger than what they designed the code to handle.

[Lalit]

Is there a reference that would indicate what the maximum count limit is? Would help with converting the overflow into a count. I've tried the samtools manual at sourceforge with no luck.

[sdriscoll]

If you can read code that seems to be the best reference for samtools. I can take a look unless somebody chimes in with a real answer. Exactly how big is this bam file anyway? Even with a 32 bit unsigned int you should be able to count way higher than any total number of reads I've ever heard of in an experiment.

[Lalit]

Its huge. The folks who did the sequencing said to expect well over 2E9 reads. (yep -- billions...)

[sdriscoll]

You could try this one...

samtools view file.bam | wc -l

See if that fails.