Need for improved methods for predicting genetic potential in animals
698 MLA says that there was a need for improved methods for predicting genetic potential in animals for numerous quantitative traits which were known to be desirable. Further, it points to the fact that the specification states at [0022] that "three different experimental approaches have been used with limited success to identify genes, chromosomal regions or DNA markers that account for a large proportion of the genetic variation observed in economically important traits in livestock species" and that the results "have not been widely utilized to date because they do not account for enough of the total genetic variation to allow accurate prediction of an animal's performance for a specific trait." MLA says that this accords with the evidence of the experts that the approaches at the time were of limited value for predicting an animal's performance with respect to the majority of quantitative traits.
699 MLA says that by the late 1990s/early 2000s, there was a lot of interest in using molecular genomics in cattle breeding and that was where research and the industry was heading. It contends that ultimately the evidence of all experts was that before December 2002, DNA markers "were the way to go".
700 MLA also says that in the review published in February 2001 of Andersson L, "Genetic Dissection of Phenotypic Diversity in Farm Animals" (2001) Nat Rev Genet 130 (cited in Wiggans GR, Cole JB, Hubbard SM and Sonstegard TS, "Genomic Selection in Dairy Cattle: The USDA Experience" (2017) 5 Annu Rev Anim Biosci 309), the author said that "[i]t is likely that large-scale marker analysis will be used routinely, as soon as the cost for genotyping has been reduced by a factor of around ten" (at 136 and 137, second column, second last line). The review also said that "[l]inkage disequilibrium mapping will be a very powerful approach for mapping and finding trait loci in domestic animals once dense SNP maps become available and the cost for genotyping has been reduced such that genome scans using thousands of SNPs can be done" (at 137, first column, second full para). The review also observed that "[i]t is also only a matter of time before initiatives will be taken to sequence the genomes of farm animals. This will most probably be carried out using a whole genome shotgun approach …" (at 137, second column, first full para). MLA says that this review was generally consistent with the views expressed by Dr Sonstegard in his 2001 review of dairy cattle genomics (Sonstegard TS, Van Tassell CP and Ashwell MS, "Dairy Cattle Genomics: Tools to Accelerate Genetic Improvement?" (2001) 79 J Anim Sci E307).
701 Further, MLA points to the fact that prior to the priority date, Professor Taylor had set up Genomic FX to do something which was similar to what is described in the 253 Application. It asserts that his company's aim was to develop a simple diagnostic test for genotyping animals for the purpose of improving their genetic potential. His plan was to scan the genomes of thousands of animals and to identify SNPs evenly spaced throughout the genome and find SNPs that correlated with useful traits. It is said that his company did not proceed with the plan because it was forced to "shut its doors". MLA says that the only difference between the work of Genomic FX and the 253 Application that Professor Taylor could identify was that he was looking at within family QTLs, not the whole bovine genome. But MLA says that the only difference between the within family QTL approach and the genomic selection approach is that GS looks at markers across the whole genome, whereas the within family QTL approach only looks at segments of the genome. Otherwise, so MLA contends, Professor Taylor's proposed approach before the priority date was the same as that of the 253 Application.
702 I would make the following observations in response to these contentions.
703 It is not in dispute that the state of the art as at the priority date was that there were a variety of methods being developed and used to determine the genetic potential of bovines for economic traits, including progeny testing, the candidate gene approach and QTL mapping.
704 It is also not in dispute that as at the priority date, traditional phenotypic selection techniques such as progeny testing were commonly used in practice and that those techniques were sophisticated and being continuously improved. Further, as at the priority date there was limited use of molecular genetic techniques. And at that time the laboratory use of molecular genetic techniques was directed more to the candidate gene approach and the within family QTL mapping approach.
705 As I have previously summarised, the candidate gene approach is based on knowledge of the gene's biological function and its relationship to the trait of interest. It is typically based on knowledge of analogous genes in other species. As Dr Sonstegard said, it was a popular approach amongst researchers at the time to identify genes responsible for quantitative traits. Further, as I have already previously summarised, QTL mapping typically involved genotyping and trait-association experiments conducted within families to identify particular chromosomal regions associated with a trait of interest. Further, Dr Sonstegard also gave evidence that when the candidate gene approach was combined with QTL mapping, the approach was referred to as a "positional candidate gene approach". The search for candidate genes was limited to a specific region of the genome that was known to be associated with a trait (i.e. the QTL).
706 It is also not in dispute that as at the priority date neither QTL mapping nor the candidate gene approach involved the use of genome wide dense markers. Such approaches were rather focused on identifying the location of a causative gene(s) within a specific region of a chromosome.
707 Now what I have just said was reflected in the research work at the time, indeed even after the priority date, being conducted by the experts who were called before me. Professor Goddard's publication closest to the priority date involved the fine mapping of a QTL locus for twinning rate in cattle which had all bulls genotyped for 15 markers located on a specific region of chromosome 5, but did not disclose the construction or use of a dense genome-wide panel of markers or map (see Meuwissen T, Karlsen A, Lien A, Olsaker I and Goddard ME "Fine Mapping of a Quantitative Trait Locus for Twinning Rate Using Combined Linkage and Linkage Disequilibrium Mapping" (2002) 161 Genetics 373 (accepted for publication on 11 February 2002 and published in May 2002)). Further, Professor Visscher's work in bovines until at least 2005 predominantly involved family-based linkage studies using microsatellite markers. Further, Dr Sonstegard was involved in the fine-mapping and sequencing of a QTL on bovine chromosome 6 and identified osteopontin as a candidate gene affecting milk production in or around 2005, which research was published in the Proceedings of the National Academy of Sciences in 2005 (Schnabel RD, Kim JJ, Ashwell MS, Sonstegard TS, Van Tassell CP, Connor EE and Taylor JF, "Fine-Mapping Milk Production Quantitative Trait Loci on BTA6: Analysis of the Bovine Osteopontin Gene" (2005) 102 PNAS 6896). Further, in evidence before me reference was made to a well-known study in 2001 in which the positional candidate gene approach was successfully used to identify a causative polymorphism in a gene called DGAT1 which accounted for about 30% of the variation in particular milk traits (Grisart B, Coppieters W, Farnir F, Karim L, Ford C, Berzi P, Cambisano N, Mni M, Reid S, Simon P, Spelman R, Georges M and Snell R, "Positional Candidate Cloning of a QTL in Dairy Cattle: Identification of a Missense Mutation in the Bovine DGAT1 Gene with Major Effect on Milk Yield and Consumption" (2001) 12 Genome Research 222). Further, Professor Plastow gave evidence about his use of the candidate gene approach on the melanocortin receptor 4 gene (MC4R), published in 2000 (Kim KS, Larsen N, Short T, Plastow G and Rothschild MF, "A Missense Variant of the Porcine Melanocortin-4 Receptor (MC4R) Gene is Associated with Fatness, Growth and Feed Intake Traits" (2000) 11 Mamm Genome 131). The MC4R candidate gene explained variation in a polygenic trait, being growth and feed intake. And according to Professor Plastow, although this gene explained a very small amount of the variation in the trait, the ability to identify variants through the use of markers was commercially very valuable. Further, I note for completeness that Professor Plastow later discussed in a 2005 article the MC4R gene, the use of functional genomics studies to identify candidate genes and also high-density marker approaches (described as Phase 3), none of which was inconsistent with his evidence before me as to what was relevantly known or expected as at the priority date (van der Steen, H, Prall G and Plastow GS "Application of Genomics to the Pork Industry" (2005) J Anim Sci 83:E1).
708 Now MLA has made reference to Professor Taylor's involvement in Genomic FX prior to the priority date. But the described objective of Genomic FX was to use within family QTL mapping with fine marker mapping to identify predictive variants or positional cloning, rather than to scan the genomes of thousands of animals and to identify SNPs evenly spaced throughout the genome. And in any event such work did not proceed. Now I accept that positional cloning was one of the methods being used in laboratories before the priority date, but this method was not a whole of genome approach.
709 Further, the evidence of Professor Taylor and Dr Sonstegard was that although there was interest in genetic markers, progeny testing was still a very common technique in the industry as at the priority date.
710 Further, and as I have endeavoured to explain previously, in contrast to the candidate gene and QTL mapping approaches, a genome wide approach does not require identifying any particular causal gene or region of the chromosome. And the possibility of using genome wide approaches in livestock breeding was theoretical as at the priority date, with it being unconfirmed whether such approaches could successfully be applied in cattle. Further, the tools required such as a genome-wide marker map, were not available. And the absence of a sufficient number of markers was a technical hurdle. Further, it was speculation whether a SNP chip would become available.
711 As at the priority date, to apply a genome wide approach it would have been necessary to identify a large number of genetic markers evenly spaced throughout the genome. And to design and carry out the necessary experiments would have presented many challenges, particularly when as at the priority date, work on sequencing the whole bovine genome had yet to commence. As Dr Sonstegard gave evidence of, which I accept, the challenges included the need to sequence and assemble a large proportion of the bovine genome, the need to identify a large number of informative markers, the need to determine the relative position of those markers, the need to generate a genome-wide panel of relatively evenly spaced markers, and the need to identify those markers that are associated with a trait. I would just interpolate at this point that MLA spent a considerable time attacking Dr Sonstegard's evidence by reference to observations made in articles that he had co-authored which discussed in part genomic selection and the problems involved and resources required. But I do not consider that MLA established at all that Dr Sonstegard's evidence was not reliable on what was known or expected as at the priority date on such matters.
712 Further, on the evidence before me, it is not seriously in doubt that as at the priority date, the relevant work was technically difficult, expensive and time consuming. The then existing techniques were inadequate. They did not provide the required specificity for SNP identification, and certainly not on a large scale. Further, SNP chips were not developed until the mid to late 2000s. I would note that they were not used by the inventors of the 253 Application. Further, as Professor Plastow explained, the necessary high throughput assay technology to screen multiple bovine markers was not developed until well after the priority date.
713 Further, the preponderance of the evidence also established that as at the priority date, microsatellites were the most widely used genetic marker in bovines. They had various advantages over other markers, including SNPs, as various of the experts accepted. Further, fragment analysers only became available in the early 2000s to enable the rate at which microsatellites could be genotyped to be substantially increased. Contrastingly, SNPs were seldom used as a genetic marker, and as at the priority date very few had been discovered in cattle. Moreover, it was only after the priority date that a significant transition towards the use of SNP markers occurred when the BovineSNP50 SNP chip was commercially released.
714 Further, according to Professor Plastow, even by 2005 the consensus was that there were still significant hurdles to be overcome before a genome wide approach could be utilised in the breeding selection and management of cattle.
715 Further, I would also note that Professor Visscher for MLA accepted that the main practical limitation to implementing a genomic selection approach as at December 2002 was the identification of a sufficient number of markers. Now although Professor Visscher said that there was no technical difficulty at the time in assaying or genotyping each individual marker, that is different to what was considered to be the main practical limitations that I have outlined. In any event, this is a reference to genotyping individual markers, rather than a genome-wide panel of markers. Now Professor Visscher accepted that the information in Vignal (Vignal A, Milan D, San Cristobal M and Eggen A, "A Review on SNP and Other Types of Molecular Markers and Their Use in Animal Genetics" (2002) 34 Genet Sel Evol 275) was consistent with his understanding of what was known prior to December 2002 about the use of genetic markers in animal genetics. But in my view Vignal does not assist MLA. Vignal does not describe the use of genomic selection as the direction of research in 2002. And I note that it does not cite the Meuwissen paper; MLA asserted that Meuwissen was well known at the time. Vignal does not suggest producing a genome wide map of evenly spaced markers. And it does not suggest how such a map could be constructed and applied. Further, Professor Visscher accepted the reference to "high densities of markers" being needed was in the context of a fine QTL mapping approach. It is well apparent that the expressed desire for more markers within the region of a QTL is different from a genome wide dense marker map necessary to carry out a GWAS or genomic selection. The purpose of identifying more markers for a fine QTL mapping approach is to more accurately locate the causal gene. But contrastingly, the GWAS and genomic selection approaches do not involve determining the identity or location of any causal mutation.
716 Further, what is apparent from Vignal, consistently with the expert evidence led before me, is that at the priority date, to the extent that genetic markers were being used, they were microsatellites. Now it might have been accepted that in the future SNPs might produce equivalent information, but that uncertainty does not suggest any expectation that SNPs could be used successfully (as claimed in the 253 Application). It is convenient to set out some extracts from Vignal.
717 Vignal, which was published in 2002 (after 8 March 2002 (the date of acceptance)), stated in section 2.2:
What is the reason for the increasing popularity of SNPs, whereas in terms of genetic information provided, as simple bi-allelic co-dominant markers, they can be considered as a step backwards when compared to the highly informative multi-allelic microsatellites? Are we not only putting a new name on what has just been considered until now as a common polymorphism and originally studied as RFLPs? In fact, the more recent SNP concept has basically arisen from the recent need for very high densities of genetic markers for the studies of multifactorial diseases, and the recent progress in polymorphism detection and genotyping techniques.
718 Further, Vignal from section 6.3 and in conclusion stated:
Several approaches can be taken for fine QTL mapping, such as increasing the number of meiosis events by increasing the size and/or the number of families for genotyping, selecting recombination events in recurrent backcrosses, using advanced intercross lines (AIL) or performing linkage disequilibrium and haplotype-based studies in outbreed populations. However, whatever the approach taken, high densities of markers will be needed. In some instances, when the populations studied are closely related, even the microsatellite markers may not be heterozygous for the F1 animals. Also, for some species, such as chickens, the density of microsatellites will be low.
Testing of candidate genes and candidate polymorphisms in exons, promoters or other important regions such as splice sites, promoters or other regulatory regions, will have to be done using the SNP approach, since this will be the most common polymorphism and the more likely responsible for phenotypic variation.
When testing for the association between complex phenotypic traits and candidate loci, single-loci SNP analyses present a loss of information due to the bi-allelic nature of the markers, as compared to the multi-allelic microsatellites. However, by performing haplotype frequency estimations over several SNPs from a locus, this can be overcome and even possibly improved, due to the fact that SNPs will more often be close to the site responsible for the variation than microsatellites.
7. CONCLUSION
Although in a strict molecular sense, SNPs are just what has been previously known as base substitutions, the fact of naming molecular markers by this acronym meaning single nucleotide polymorphism, is an indication of the new importance that this type of polymorphism has in molecular genetics. Indeed, if in some instances, the lack of information due to the bi-allelic nature of SNPs is a limitation, there are cases in which they can provide valuable data on associations between specific genes or other DNA structures and phenotypes, or on population and genome dynamics.
The very high density of SNPs in genomes, usually allows to develop several of them in a single locus of a few hundred base pairs. By reconstructing haplotypes, multi-allelic systems can eventually be defined for analyses, to overcome the limitations due to the low heterozygosity of SNPs. With increasing progress being made in the molecular techniques used to produce SNP data, in the automation of allele scoring and in the development of algorithms for genetic analyses, the effort needed to produce an equivalent amount of information as with microsatellites may some day be equivalent.
719 There is little doubt that the broader use of SNPs was expected to evolve.
720 In summary, although I readily accept that as at the priority date there had always been a desire to improve breeding methods in domesticated animals, it is readily apparent that at that time traditional non-molecular genetic approaches were the most widely used. Moreover, if genetic markers were used to identify causative genes, these markers were principally microsatellites.