Exome sequencing: should we do local realignment

[yavit]

Dear colleagues,

With ever-increasing quality of NGS platforms, is it still desirable to do local realignment for HiSeq 2500 human exome reads from an Agilent Sureselect library?

Thank you

[DNATECH]

Local realignements are helpful not because of sequencing quality concerns, but because of the general difficulty to get alignments right around indels, especially towards the end of reads .
Btw, just saw this nice paper: http://journals.plos.org/plosone/art...l.pone.0075402

[Brian Bushnell]

I normally trim the first and last 5bp of reads for this reason (to avoid miscalled indels at the tips), after the reads are aligned. If you don't realign around indels, I recommend this. Might be a good idea even if you do realign; even with realignment, there are some features that you cannot resolve, since there is not enough information at the read tips.