1. rare Mendelian diseases versus common multi-factorial diseases e.g., cystic fibrosis is one of the most common life-shortening childhood-onset inherited diseases in the United States, affecting 1 in 3900 births; one of every 31 individuals is a carrier of the recessive disease allele e.g., in the United States, the lifetime risk for developing cancer is slightly less than 1 in 2 for men and slightly more than 1 in 3 for women; although there are subclasses of cancer like early onset breast cancer that obey Mendelian rules, they make up a negligible fraction of the overall disease

2. 1980: DNA markers are the key to identifying Mendelian disease genes Botstein D, White RL, Skolnick M, Davis RW. 1980. Am J Hum Genet 32: 314-331 the fact we cannot experiment on people does not preclude us from doing genetics; all we need are more DNA markers to differentiate individuals; the markers themselves need not cause the disease; they need only be sufficiently close to the gene that does; markers can take many forms including RFLPs, microsatellites, SNPs, etc. (a) disease cosegregates with marker A; (b) disease cosegregates with marker B prior to a recombination event, and marker C after power to localize genes is reliant on having sufficient number of recombinations (a)(b)

3. 1989: successful cloning of CFTR gene responsible for cystic fibrosis Rommens JM, …, Tsui LC, Collins FS.1989. Science 245: 1059-1065 the  F508 mutation is one of over 1500 eventually found in this gene but it is still the most important; this was the first disease gene identified by its chromosomal position instead of by its hypothesized function; Huntington’s was mapped before CF but the unexpected nature of that mutation (triplet repeat) took longer to solve

4. Online Mendelian Inheritance in Man currently lists 2284 phenotypes whose molecular basis is known the success of what came to be known as positional cloning was a tribute to an admission of ignorance; we did not know enough human biology to guess the likely gene for a disease so we focused instead on determining where the gene was on the chromosome; for the overwhelming majority of cases, the answer turned out to be a completely unknown gene that no scientist had hypothesized

5. 1990-6: birth and death of sib pair analysis for linkage based studies of common multi-factorial diseases Risch N, Merikangas K. 1996. The future of genetic studies of complex human diseases. Science 273: 1516-1517 linkage analysis had been successfully used to find genes for Mendelian diseases; in 1990, Risch popularized a method (sib pairs) to find genes for complex multi-factorial diseases; that method failed and they wanted to propose a different method that would be more powerful association studies were to be performed on functional polymorphisms for as many candidate genes as technically feasible, the entire genome if need be, regardless of how impractical that was; at least the number of patients would no longer be a limiting factor

6. past, present, and (near) future genetic studies of human diseases the biggest change from before is we are now doing these studies on the general population instead of rare families we must study common diseases in the general population because rare mutations that cause Mendelian subcategories of disease are not typically responsible for those diseases in the general population the problem however is that without families we lose statistical power and must compensate by gathering more data

7. population bottleneck, subsequent recombination, linkage disequilibrium although the parameters and details of the human population bottleneck are still not settled, the order of magnitude estimates are that our species collapsed to 15,000 individuals 70,000 years ago; assuming few new mutations the only thing that would have happened since that time is recombination, and we can model any particular individual’s genome as a mosaic of segments from these 15,000 ancestral genomes mutation is the red ; chromosomal stretches derived from common ancestor are yellow; the new stretches due to recombination are blue

8. we need not test all the functional polymorphisms, just enough markers to be within linkage disequilibrium linkage disequilibrium (LD) is the non-random association of two alleles on adjacent loci; there are many reasons why this might happen, but for the HapMap, the assumption is that human variation is intrinsically limited because of the recent population bottleneck

9. common-disease-common-variant versus common-disease-rare-variant CDCV hypothesis: a few common allelic variants account for most of the genetic variance in disease susceptibility Reich DE, Lander ES. 2001. On the allelic spectrum of human disease. Trends Genet 17: 502-510 CDRV hypothesis: a large number of rare allelic variants account for the genetic variance in disease susceptibility Terwilliger JD, Weiss KM. 1998. Linkage disequilibrium mapping of complex disease: fantasy or reality? Curr Opin Biotechnol 9: 578-594 for complex reasons having to do with human population history, linkage disequilibrium would only work in diseases where the CDCV hypothesis is valid; the best justification for the HapMap was that one common variant has more public health impact than many rare variants, so it makes sense to find these first

10. multiple rare alleles contribute to low plasma HDL cholesterol levels for 128 individuals with low plasma HDL-C, 21 (16%) had variants not present in the high HDL-C group; conversely, only 3 (2%) of individuals with high plasma HDL-C had variants not present in the low HDL-C group (P < 0.0001); Cohen JC, …, Hobbs HH. 2004. Science 305: 869-872 3 non-synonymous variants in 3 out of 128 outlier samples 15 non-synonymous variants in 21 out of 128 outlier samples HDL-C levels candidate genes ABCA1, APOA1, LCAT

11. International HapMap Consortium http://www.hapmap.org/thehapmap.html.en International HapMap Consortium. 2005. Nature 437: 1299-1320 phase I genotyped common SNPs of frequency greater than 0.05 in every 5-kb interval for 269 individuals from 3 populations in Africa (YRI), Europe (CEU), and Asia (CHB+JPT) solid line represents ENCODE region data, dashed line represents neutral model with constant population size and random mating and no ascertainment biases

12. 7 tag SNPs capture all the common variation in a locus on chromosome 2 left plot shows the 7 haplotypes and their respective counts, with colored circles indicating SNP positions where a haplotype has the less common allele; groups of SNPs captured by a single tag SNP (r2  0.8) using a pairwise tagging algorithm have the same color; right plot shows the SNPs mapped to a genealogical tree relating the seven haplotypes for this region

13. Wellcome Trust Case Control genome wide association studies Wellcome Trust Case Control Consortium. 2007. Nature 447: 661-678 genotype 500,000 SNPs from HapMap in a British population of 2,000 affected individuals for each of 7 major diseases, with another 3,000 shared individuals for control of the 14 variants for which there was a strong prior evidence of association to the studied diseases all but two (APOE and INS) were reproduced by this analysis

14. genome wide scan in seven diseases; y-axis represents statistical significance using -log 10 of a p-value the chromosomes are shown in alternating colors; significant SNPs with p-value <1  10 -5 are in green

15. a doubling in relative risk for a disease is not as bad as it sounds Couzin J, Kaiser J. 2007. Science 316: 820-822 with the notable exception of macular degeneration there is only a doubling in relative risk; the 120% increase in relative risk for inflammatory bowel disease only bumps the absolute risk from 0.5% to 1.1%

16. but gene therapy remains elusive 19 years after the cystic fibrosis gene Jesse Gelsinger (June 18, 1981 to September 17, 1999) was the first person identified as having died in a clinical trial for gene therapy. He was only 18 years old. Gelsinger suffered from ornithine transcarbamylase OTC deficiency, a disease of the liver whose victims are unable to metabolize ammonia, a byproduct of protein breakdown. Gelsinger was injected with adenoviruses containing the corrected gene in the hope that it would manufacture the much needed enzyme. He died four days later, having suffered a massive immune response, triggered by the viral vector used to transport the gene into his cells. This led to multiple organ failure and brain death. Food and Drug Administration investigators concluded that scientists involved in the trial, including lead researcher Dr. James M. Wilson (University of Pennsylvania), broke several rules of conduct: (a) Inclusion of Gelsinger as a substitute for another volunteer who had dropped out, despite his having high ammonia levels that should have led to his exclusion from the trial, (2) Failure by the university to report that 2 other patients had experienced serious side effects from the therapy, (3) Failure to mention the deaths of monkeys given a similar treatment, as should be been done for the informed consent. The university paid the parents an undisclosed amount.

17. http://www.genetic-future.com/2008/03/why-do-genome-wide-scans-fail.html http://www.genetic-future.com/2008/03/why-do-genome-wide-scans-fail.html Alleles with small effect sizes: To separate true signals from noise, researchers have to set an exceptionally high threshold that a marker needs to exceed before it is acceptable as a likely disease-causing candidate. By increasing the numbers of samples in their disease and control groups, researchers will steadily dial down the statistical noise until even disease genes with small effects stand out above the crowd. However, the logistical challenge of collecting a large number of carefully-ascertained patients will always be a serious obstacle. Rare variants: Our catalogue of human genetic variation, i.e. the HapMap, is largely restricted to common variants, since rare variants are much harder to identify. The instrumentation has restrictions on how many different SNPs one can analyze with a single chip. Everyone agrees that some non-trivial fraction of the genetic risk of common diseases will be the result of rare variants, especially as the latest results in a variety of diseases failed to provide unambiguous support for the CDCV hypothesis. The problem is not so much the costs of sequencing itself, as that is plummeting due to massive investment in rapid sequencing technologies, but rather the interpretation of the resultant data. Population differences: Markers that are associated with disease in one population can never be assumed to show the same associations in other human groups. This will be especially true for rare variants. The more difficult challenge will be in collecting large numbers of ancestry- homogeneous samples of validated disease patients and healthy controls.

18. http://www.genetic-future.com/2008/03/why-do-genome-wide-scans-fail.html http://www.genetic-future.com/2008/03/why-do-genome-wide-scans-fail.html Epistatic interactions: Most genetic approaches assume that genetic risk is additive, in other words, that the presence of two risk factors in an individual will increase risk by the sum of the two factors. There is however no reason to expect that this will always be the case. Epistatic interactions, in which combined risk is greater (or less) than the sum of the risk from individual genes, are difficult to identify by genome scans and even harder to untangle. Copy number variation: CNVs are now known to account for a substantial fraction of human genetic variation, and have been shown to play an important role in gene expression variation and in human evolution. It seems highly likely that CNVs will be responsible for a non-trivial proportion of disease risk, but only time will tell. Epigenetic inheritance: Although epigenetic inheritance does occur, the degree to which it is influencing human physical variation and disease risk is essentially unknown. It needs to be established that epigenetically inherited variations actually contribute to a non-trivial fraction of human disease risk before we can include them in our systematic scans. Disease heterogeneity: Lumping patients with fundamentally different conditions into a single patient cohort is a recipe for failure, even if there are strong genetic risk factors for each of the separate conditions, since each will be drowned out by noise from the other. The geneticists cannot fix this problem. It will take a combined effort of clinicians and biomedical researchers to stratify the disease into useful subcategories.