The use of MAPH in assessing copy number changes at new loci requires the generation of new probe sets, which can be a lengthy and time-consuming process if individual amplicons need to be designed and cloned. Since many of the most likely candidates for disease association are in regions of the genome that also contain repeat elements, the positioning of single-copy probes can be problematic. Therefore efficient screening of regions for copy number variants by MAPH has necessitated the development of QuadMAPH and with it, the introduction of new cloning systems and a new method of generating probes.
Previously, MAPH probes have been made by designing and cloning individual PCR amplicons [20
], which is not always appropriate for covering large genomic regions, or if numerous probes are required. In order to screen regions of approximately 200 kb in size with a probe approximately every 1-2 kb, we have established an alternative approach for MAPH probe generation, by which sets of probes are prepared by shotgun cloning of random short fragments from a BAC clone covering the region of interest. This method of producing MAPH probes is both efficient and cost-effective with respect to the number of probes generated and can be easily and rapidly applied to new loci which are not covered by commercially available technologies. An important step in the random generation of fragments is the choice of restriction enzymes in the initial BAC DNA digest, which dictates the number of usable MAPH probes produced. The use of double stranded linkers allows simultaneous PCR amplification of all fragments generated by this digestion. Digested BAC DNA was subject to two rounds of size-selection, which resulted in a reasonable distribution of fragments in the size range suitable for MAPH probes and was also necessary to offset any potential bias towards small fragments in the subsequent cloning and transformation steps.
Since the clones were generated randomly, unsurprisingly there was a degree of redundancy of fragments between vectors, but this allowed flexibility in the choice of probes for each set, and also produced a reserve of probes that could be used to improve on resolution or to confirm any potential copy number changes. In fact, once a set of clones has been generated, these can be stored in their 96-well microtitre plates at -20°C to -80°C thus providing a collection from which to make additional MAPH probe sets with minimal additional labour. Despite the high density of repetitive elements (SINEs) at the MSH2 locus, this approach generated good probe coverage across the region, as illustrated in Figure . If probe density is not sufficient to interrogate a specific region, then gaps can be filled by custom design of a small number of supplementary probes. Custom probes can be cloned into any one of four vectors to fit into size gaps in the profile for the vector chosen.
All four vectors can be multiplexed in any combination or used in MAPH on their own. Therefore, using four cloning systems allows the generation of four independent MAPH probe sets and subsequently a substantial increase in the number of targets that can be interrogated in a single test. The number of loci that can be analysed at once is determined by the number of probes in the probe set. Theoretically, the number of probes per set is only limited by the resolution and the number of fluorophores detectable after capillary electrophoresis, and in the case of generating probes by a random shotgun cloning approach, the size range of probes produced in the initial digest. Whilst our initial MSH2 QuadMAPH set consisted of 110 probes, additional probes can be included. In this work, by choosing an initial copy number screening resolution of a probe every 1-2 kb, the boundaries of any copy number variants discovered could then be defined within the range of long PCR, and junction-fragment analysis can be used to establish a rapid, specific PCR assay for the variant.
MAPH has previously been demonstrated as an accurate and precise assay for measurement of copy number [20
]. After extensive experimental work dissecting and analysing factors affecting MAPH's accuracy and robustness, improvements have been made in the working method. PCR cycling modifications, and adherence to a strict wash protocol and wash solutions appropriate to the %GC content of the probe set, have reduced copy number measurement error and improved the reliability and reproducibility of the assay. As a consequence of this, the standard deviations reported here are considerably smaller than those reported by Hollox et al
]. Associated with this is a substantial reduction in the predicted incidence of false positive and false negative results. The incorporation of control probes within a probe set, such as X and Y linked probes and a non-human probe, allow the user to assess non-specific probe binding and adjust washing conditions as necessary, as well as confirm that any copy number changes can be detected accurately based on the responsiveness of the control probes.
By comparison of data obtained by simplex MAPH with those by QuadMAPH, we have also demonstrated that multiplexing four probe sets is achieved with no ensuing loss of measurement accuracy, as indicated by the false positive and false negative predictions. Although MLPA has been successfully used in a two-colour format [23
] these studies examined totals of 28 and 45 probes respectively, and we are not aware of successful use of MLPA at probe multiplicities approaching those in this report. Since MLPA also depends on pairwise ligation of half-probes, there may be a very large number of opportunities for MLPA using a large number of probes and multiple primers to create artefactual products; by contrast, MAPH probes in different vectors are not expected to interact with one another, and we have shown here that 110 probes perform well together. In summary, we believe QuadMAPH has the capacity to screen large numbers of loci simultaneously on a large number of samples.