Biology of Genomes Key Notes

Dr. George Davey Smith *Bristol): Key Note 1

• Epidemiologist
• was claimed that vitamin E reduced heart disease (observational studies), studied carefully in RCTs NO EFFECT
  • lesson of confounding factors
• Mendialian randomization
  • no reverse causation in genetics, instrumental variable.
  • Mendelian mutation effects something which affects something else.
• examples C-reactive protein and interlukin 6 associate
• inlt-6 and fibrogen interact. and c-reactive and fibrogen interact
  • all effect heart disease
  • can’t get stat effect
  • Ln C-reactive protein robustly linked (explains > 1% of the variance)
  • genotype raised concentration of CRP – no effect. Ditto for Fibrongen
  • for Il6, high Il6 is actually linked to heart disease.
• Mendilian randomization as analogous to a randomized control trial
• Mendilian case – meiosis randomizes SNP linked to Se
  • SNP corresponds to different genetic lvels of selenium
  • both true RCT and Mendelian approach give same results on 20,000 + case/control study
• another example – low body fat linked to lung cancer (because smoking reduces BMI). Mendilian randomization study clearly removes this confounding.
• Multiphenotype Mendilian Randomization (MR)
  • lipids and CHD as an example. lipid from MR not singfincant
  • adjusting HD-L lowering risk of heart disease looks good, true effect
• Limitations: reintroducing confounding via pleiotropy
• Egger-regression: regress effect of SNP on effect of phenotype, can test existence of pleiotropy (from x-intercept) and still measure effect from slope. – address limitation of pleiotropic effects
• interact instrument with a second variable:

alcohol consumption

• ALDH2 mutants: males homozygous WT drink more than hets and homo don’t drink. women don’t drink.
• drinking alcohol actually increases blood pressure — males who don’t drink have lower blood pressure (no effect in women, allele)
• with genetic variant the mimics drug effect, can efficiently / cheaply conduct randomized trial.

Questions

• non-single gene loci are a problem, but with enough data with independent combinations of these, it can be addresed

Francis Collins

• (PhD Yale, MD NC)
• what you may not know:
  • man who led the Human Genome mapping / sequencing
• NIH director since 2009

Reflection from HGP to Precision Medicine

• first time back since 2011
• Human genome 1990-2003
  • challenge to public project from private industry
• not in the post-genome era. We’re in the genome era.

Major advances

• tumor cancer genomics
• explosion of human microbiome
• chromatin open or closed
• GTEx (3 papers today in Science) + 3 other elsewhere:
• the Big data problem: BD2K
  • big data to knowledge project (100 million per year) BD2K
  • NIH’s 6-year iniatitive
  • NCBI 10 Tb/day, 40 Tb/day downloads, 3Tb/day interactive (exponential growth)
• future of National Library of Medicine
  • active working group

The Case for Precision Medicine: ‘Timing is Everything’

• form some large scale prospective cohort
  • cost per human genome 1-5K in < 1 day
  • number of smart phones
• announced in State of Union Address ‘precision medicine initative’
  • supposed to start this October
• what is precision medicine?
  • fit he patient (not fully new, e.g. glasses)
  • most medical things are given for the ‘average patient’ (if for any scientific reason at all)
• why now?
  • electronic health records
  • wearable medical sensors
  • genomics
  • metabolomics
• what’s needed now?
  • rigorous research program (need to recruit people)!

Vision

• personal / precision medicine advanced the furthest in cancer
• patient partnerships, Elect. Health. Rec (EHR).
• president proposes budget increase of 215 million (mostly through NIH, 70 for cancer, 130 for cohort).
• reasonable chance of being passed by congress.
• ‘liquid biopsies’ (circulating tumor DNA)
• other new technologies? Multi-therapy approaches?
• ID mass with liquid bioposy showing tumor risk mutation.
• Longer term: pilots to build up cohort to 1 million+ volunteers
  • already millions involved in existing NIH funded longitudal studies

Cohort

• data driven cohorts – psychaiatric diseases for example clearly lack molecular based clustering / appropriate for
• human knock ID – nature’s solution of diseases resiliance. ID protective genetic factors (and other factors)
• Pharmacogeneomics – over 100 drugs list information about genetic influences on the label (largely being ignored now)
• Annual physical exam (not so much evidence this is useful).
• “Make no little plans, they have no magic to stir men’s blood and probably themselves will not be realized. Make big plans; aim high in hope and work” – Daniel Burnham

Questions

• 23 and me approaching 1 million
• what’s the role of basic research in this initative?
  • 53% of NIH to basic science
  • this will be more of a clinical / applied bent
• provision for training doctors?
  • that’s a challenge