Bamutils/ClipOverlap

Question

[chrisk]

We are using clipOverlap from bamutils for read overlap removal (post alignment) from our BAM files.

The clipOverlap (https://genome.sph.umich.edu/wiki/BamUtil:_clipOverlap) manual states :

"It also checks strand. If a forward strand extends past the end of a reverse strand, that will be clipped. Similarly, if a reverse strand starts before the forward strand, the region prior to the forward strand will be clipped. If the reverse strand occurs entirely before the forward strand, both strands will be entirely clipped. If the --unmapped option is specified, then rather than clipping an entire read, it will be marked as unmapped."

We understand the forward strand extending past the end of the reverse strand, but are unable to visualise what the statement in bold is specifically describing? How does the reverse strand start before the forward strand, but then clip this region prior to the forward strand - it sounds as though it is not the overlapping segment?

We also observed the third rule in action where discordant reads with no overlap were entirely clipped. Any thoughts as to why they would perform this action on reads that have no overlap? We have a minimal amount, but are interested to understand their theory to better understand the software in general.

Once again thank you for your time! ( This is a repost of a question in an older thread - accidentally doubled up with multiple questions.)

[GenoMax]

it sounds as though it is not the overlapping segment?

That is my interpretation as well.

Any thoughts as to why they would perform this action on reads that have no overlap?

Since this tool is supposed to clip extra sequences beyond the point of overlap of two reads they are probably removing reads that do not satisfy the main operational requirement of the tool. If you want to keep such reads then you will need to use --unmapped option.