Fig. 6

Merging paired-end reads. Paired-end reads with identical prefix at both ends can be considered potential duplicates. The same clustering strategy used to identify potential duplicates in single-end reads can also be used for paired-end reads. In this case, paired-end reads need to be merged as represented in the figure. A sequence representative of a pair is obtained by merging the prefixes and the suffixes of both forward and reverse read. With N the length of the read sequence and p length of the prefixes, the new sequence consists of 2·N nucleotides and is represented by a prefix of 2·p nucleotides