Limitations and Improvement of Constructing Long Paired-end Libraries

Author(s): Cheng, Jan-Fang; Froula, Jeff; Ewing, Aren; Peng, Ze | Abstract: Long insert size paired-end (PE) sequences play a key role in de novo assemblies of complex genomes. There is no reliable method, however, for generating 30Kb or longer insert size PE libraries that is suitable for 454 or Illumina sequencing. In the attempt to understand the limitations of constructing greater than 30Kb PE libraries, we have purified 18, 28, 45, and 65Kb sheared DNA fragments from yeast and circularized the ends using the Cre-loxP approach described in the 454 Titanium long paired-end library protocol. With the increasing insert sizes, we found a general trend of decreasing library quality in several areas. First, the portion of redundant pairs increases from 18percent to almost 40percent of the total pairs when the average fragment size increases from 18Kb to 45Kb. Second, the contamination of short insert size pairs (l10Kb) increases as the targeted fragment size increases. Our explanation for these quality changes is that as the targeted fragment size increases, the efficiency of the Cre joining two ends of the targeted fragments decreases. Therefore the complexity of the long pairs decreases. As a result of the PCR amplification, the redundancy of the targeted long pairs increases. On the other hand, the end joining of contaminated short fragments from the size selection gel is much more efficient than the targeted fragments. As the targeted fragment size increases, the population of the short pairs becomes more prominent. Third, we have also observed the chimeric rate increases with the targeted insert sizes. It is conceivable that as the efficiency of joining both ends of large fragments decreases, the chance of joining ends between DNA molecules increases, hence the increased chimeric rate. We have been testing several changes in the protocol. We now run two gels to size select targeted fragments to remove contaminated short fragments from the first gel. We have also tested the decrease of DNA concentration in the Cre recombination reaction to reduce the chance of inter-molecular joining. We will discuss the quality improvement of these changes.