We are currently using V4 Illumina Sequencers. I am trying to estimate optical duplicates and to do so I have recently upgraded PicardTools to V 1.122 to enable readname abilty. The generated metrics file now outputs a non-zero value for READ_PAIR_OPTICAL_DUPLICATES which seems reasonable.
When I change the option OPTICAL_DUPLICATE_PIXEL_DISTANCE in the PicardTool MarkDuplicate I get differing duplicate numbers of course. For example if the option is left at default (100pixels) the tool returns 1111 optical duplicates for a particular BAM. When i change the option to 10pixels the same BAM is shown to have 60 optical duplicates.
My question, what is the ideal pixel distance to set OPTICAL_DUPLICATE_PIXEL_DISTANCE to for this V4 generation of Illumina sequencer? Is the default good enough?