I have a Fastq dataset obtained from Miseq with 24 million reads. When I use the barcode splitter tool to split them to 3 groups, I found that the matched reads are only about 50k, 9k and 900 reads. There are 99% reads are unmatched.
But I manually check the first megabyte (about 4500 sequences), each of three barcode gets more than 1000 matched reads.
I have no idea with this problem. Any suggestion is welcome.