I have a question about how picard mark duplicates: in one set of duplicated reads, would all of them be marked with Flag 1024, or only the ones for removal? In another word, would picard distinguish the "best" read from the duplicated pool of sequences by not giving it a flag?
I tried to go over picard's source code, but I'm not familiar with JAVA, so unfortunately failed:( Thanks in advance for your help here!
Oh woo I'm sorry I didn't know Picard has this explained already. Thanks!