Adding to that - 806-515=291. You can't expect to see longer inserts than that, so you should expect all of your reads to have insert sizes shorter than read length and thus run into adapter sequence at the tails. That said, hypervariable regions are called that for a reason, so potentially some pairs may have an insert size over 300bp. It's a good idea to use a read merger that can handle both scenarios (e.g., not FLASH). Or, adapter-trim the reads prior to merging, which is a good idea regardless.
• 0 views
•
link
written
by
Brian Bushnell
Were those reads trimmed?
No, those reads weren't trimmed.
all of your R1 and R2 reads have this type of orientation ? or only a subset ?
Yes, all of R1 and R2 reads have this type of orientation.
From where did you get the data? From a public database or your own custom data? If from public database share the identifier
It's our own custom data, and not from public database.
I think of having found the explanation here: http://thegenomefactory.blogspot.fr/2013/08/paired-end-read-confusion-library.html
(See "Adaptor read-through")
Are you sure you didn't just swap the R1 and R2 files?
Did you do a gel cut? When you did, were your adapter already added in the mixture? If so, you probably selected for very short fragments.