Is there a way to predict the sizes of alignment files based on the input data?

[TabeaK]

Okay, here is something I have been wondering...

For any of the popular short read aligners...I know the size of my input index...I know the size/s of my fastq files for alignment...can I calculate or estimate somehow how big the resulting sam/bam file is going to be?

[westerman]

It depends on if you are allowing and requiring a single alignment per read or allowing either an unlimited or limited number of alignments. For the first you will have a one-to-one correspondence of reads in your input to the lines in the output sam/bam. The length of the fields in the sam file are going to be roughly the same from read to read so you should be able to treat each line as the same length. As a guess, I'd say that the sam file will be twice as large as the input fastq file. The compression of the bam file is variable, of course.

If you allow multiple alignments or do not require a read to have an alignment (even as a null alignment), then it becomes impossible to say up-front what the size of the sam/bam file will be.

How is that for a "it depends" non-answer?

[TabeaK]

Pretty good actually!

I think I'll go with the: Size of SAM = size of reference + size of fastq-reads + x for now...