Sequencing of seed BACs on all chromosomes was conducted at University of Oklahoma, following which UOK, JCVI, INRA and Sanger assumed responsibility for its chromosome(s). Seed BACs were extended using BAC end information (and with reference to any available FPC map) or by BAC-filter hybridization. Candidate BACs and BAC overlaps were validated by fingerprinting and/or PCR or Southern hybridization. BACs with atleast a 2kb overlap and providing maximum extension were selected. Sequencing was conducted by standard paired-end sequencing of 3-6 kb small insert shotgun libraries.
BACs were sequenced as pools by 454 methodology using the Clone-Array Pooled Shotgun Sequencing (CAPSS) strategy (Cai et al., 2001) and it's refinement (Csuros and Milosavljevic, 2004), by pooling BACs in 10x12 array format. Each separately pooled column or row was sequenced in an individual quarter quadrant on 454/Roche GS-20 yielding 105bp average length reads.
In later stages, DNA from BACs representing the minimal tiling path for selected FPC contigs was isolated, sheared, pooled and tagged using one of the 12 initial 454 MID tags. Such libraries were pooled in groups of 12 and sequenced by 454/Roche GS-FLX yielding 210 bp average length reads. Individual FPC contigs were obtained directly by deconvolution based on the original MID tag used, followed by sequence assembly.
Majority of BACs sequenced at Genoscope were via Sanger technology while the last set was sequenced using multiplexing on 454 sequences with GS-FLX and Titanium versions. Assemblies were performed by Newbler and finishing reactions were performed for all clones, until all contigs could be ordered and oriented.
Sequencing accuracy across overlapping phase 3 BACs over all chromosomes was 99.98% and across all BAC overlaps was 99.92%.