Mega Weaver: A Simple Iterative Approach for BAC Consensus Assembly

Hierarchical genome assembly can be divided into three distinct stages: sequencing and assembling shotgun reads for each of a series of selected BAC clones; assembling the resulting fragments into BAC consensus sequences; and mapping and orienting the BAC consensus according to external positional information. We report a new approach for BAC consensus assembly that relies on iterative layouts of overlapping sequence, with no need for prior masking of repetitive sequence. The approach includes major steps of quality filtering and an iterative screening algorithm within and between clusters of overlapping BAC fragments. Each step includes numerous minor steps designed to detect false overlaps at minimal expense in true overlaps. In contrast to dynamic algorithms, our approach attempts to minimize false overlaps before attempting to form BAC consensus sequences. We show that false overlaps are reduced to a degree that final BAC consensus assembly is straightforward under a coordinate system described in the paper. Using human chromosome 22 and a range of simulation conditions, an average of 98.1% false overlaps could be removed, while 6.7% of true overlaps were not detected. The final assembled BAC consensus sequences were nearly optimal, and support the usefulness of our approach for future hierarchical sequencing projects.