International Rice Research Institute, P. 0. Box 933, 1099 Manila, Philippines

     
     A BAC library of IR64 was constructed at the Genome Mapping Laboratory at IRRI (Yang etal. 1997) with a total of 18,342 clones stored in 48 microtiter plates. Each plate is in an array of 384 wells with 16 rows and 24 columns.

Two levels of BAC DNA pools (primary and secondary pools) were prepared for PCR analysis. The primary pools indicated in capital letter (see Fig. 1), were based on the entire BAC library and were prepared with a three dimensional pooling scheme. The first dimension was plate pools with each containing 384 BAC clones from one plate. The second dimension was row pools. Each of 16 row pools contained 1152 BAC clones (24 x 48) from the same row of 48 plates. The third dimension was colunm poois with each consisting of 768 BAC clones (16 x 48) from the same column over 48 plates. The secondary pools indicated in smaliletter, were based on each microtiter plate of the BAC library. Two dimensions (row and column) pooling were made for each of 48 plates. Bacterial cells from each row (24 clones) or each column (16 clones) were separately pooled and 40 secondary DNA pools were finalized for each plate.

    Twenty two arbitrary primers were randomly selected for our experiment. In the first round surveying of the primary BAC pools, five to fifteen bands (loci) were produced from each primer. One to ten polymorphic bands with an average of 4.4 could be effectively scored. For each specific band, several secondary pools were fished out from the plates identified in the first round. Further PCR analysis was made to verify the overlapping clones (see Fig. 1). Altogether, a total of 246 BAC clones were identified with 22 arbitrary primers and confirmed by DNA fingerprinting and Southern analysis. They were dispersed in 97 loci in the rice genome. Seventeen of these loci were represented by only single clone. The other 229 clones were grouped into 80 overlapping groups. On the average, each arbitrary primer identified 11.2 BAC clones which formed 4.4 contigs with 2.5 BAC clones each (Table 1).

     Two clones (24p5 and 42m8) were found repeatedly in two different groups. The first group identified by primer AJ5 contains 4 clones (24p5, 42m8, 37pl4, and 41b8) which have the same amplified fragment of 2,000 bp. The second group detected by primer M2 contains 5 clones (24p5, 42m8, 12a3, 45d21, and 34g7) which have the same amplified fragment of 1,100 bp. Southern hybridization of these clones with 24p5 as probe proved they overlapped. We found three cases where two overlapping groups of BACs (identified by 2 different primers) were connected into a single larger contig by sharing a single common clone. The more the arbitrary primers are used, the more contigs are expected to be linked together to form larger contigs.
     AP-PCR is sensitive to many factors like the concentration of target DNA. In our research, a few new fragments amplified in the first round of BAC DNA pools were not shown in the control parent, IR64. A similar situation was also found among different pooled samples. For instance, AP-PCR using row or column pools containing 3 and 2 folds of BAC clones tended to produce no or fewer amplified fragments than from plates pools. It means that the size (number of BACs pooled) of a DNA pool is an important factor for efficient screening overlapping clones by AP-PCR analysis. In other words, the determination of size of DNA poois should be considered according to the genome size.