DOE Human Genome Program Contractor-Grantee Workshop VIII
February 27-March 2, 2000  Santa Fe, NM

46. Analysis of WUSTL's Human BAC Fingerprint Database

R. Sutherland, M. Mundt, and N. Doggett

Bioscience Division and DOE Joint Genome Institute, Los Alamos National Laboratory, Los Alamos, NM 87545

rds@lanl.gov

We have used the LANL human chromosome 16 BAC contig data to evaluate the Washington University's Genome Sequencing Center Human BAC Fingerprint Database.

WU has fingerprinted 162,272 RPCI-11 BACs and assembled them into 12,549 contigs.

LANL has identified 4085 BACs using 1106 overgos and STSs from 16 q-arm. The 16 q-arm is 45 Mb and covers 1.35% of the human genome.

For this exercise, only BACs from sections 1 and 2 of the RPCI-11 library were considered. For these sections, there are 125,979 WU's BACs within contigs and 3530 LANL mapped BACs.

The first results are a straight set-to-set comparison of the two data sets to see which WU contigs can be linked to 16 q-arm map. BACs occurring in the LANL set were used to query WU contigs. The results were as follows: 10,882 BACs from 657 contigs were identified from the WU data, 2034 BACs were in common with the LANL data. 57% of the LANL BACs could be found in a WU contig but only 18.7% percent of WU BACs in these contigs were found in the LANL set.

For the second analysis we discounted all WU contigs that contained only a single LANL mapped BAC. The results were as follows: 3,618 BACs from 245 contigs were identified from the WU data, 1622 BACs were in common with the LANL data. 46% of the LANL BACs could be found in a WU contig while 44.8% percent of WU BACs in these contigs were found in the LANL set.

For the third analysis we limited the WU set further. Only BACs that are contained in contigs that range from 2-40 members were considered; this is a 1 sigma distribution. The results were as follows: 2,766 BACs from 236 contigs were identified from the WU data, 1491 BACs were in common with the LANL data. 42% of the LANL BACs could be found in a WU contig while 53.9% percent of WU BACs in these contigs were found in the LANL set.

We believe that the LANL BAC map provides >90% coverage of the 16 q-arm and that we identified the great majority of 16 q-arm BACs from sections 1 and 2 of the RPCI-11 library. Thus, the percentages above suggest to us that there is a significant level of false overlaps in the WU BAC contigs.