1. Individual NUC genes (expression) impacts aging

1. Individual NUC genes (expression) impacts aging
2. NUC pathway (not just individual genes) impacts aging

1. Individual NUC genes (expression) impacts aging
2. NUC pathway (not just individual genes) impacts aging
3. Phenotypes of aging are related
   • parental lifespan
   • healthspan
   • longevity 90
   • longevity 99

1. Individual NUC genes (expression) impacts aging
2. NUC pathway (not just individual genes) impacts aging
3. Phenotypes of aging are related
   • parental lifespan
   • healthspan
   • longevity 90
   • longevity 99
4. Early termination-of-adult lifespan (TAL) expression as a pathway impacts aging
   • positive control

1. Individual NUC genes (expression) impacts aging
2. NUC pathway (not just individual genes) impacts aging
3. Phenotypes of aging are related
   • parental lifespan
   • healthspan
   • longevity 90
   • longevity 99
4. Early termination-of-adult lifespan (TAL) expression as a pathway impacts aging
   • positive control
5. Individual TAL genes impact aging

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.
2. Ran MR for the random sets for each of the aging phenotypes for NUC and TAL.

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.
2. Ran MR for the random sets for each of the aging phenotypes for NUC and TAL.
3. Calculated the proportion of significant findings for the null pathways.

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.
2. Ran MR for the random sets for each of the aging phenotypes for NUC and TAL.
3. Calculated the proportion of significant findings for the null pathways.
4. Generated eight null distributions (4 NUC and 4 TAL).

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.
2. Ran MR for the random sets for each of the aging phenotypes for NUC and TAL.
3. Calculated the proportion of significant findings for the null pathways.
4. Generated eight null distributions (4 NUC and 4 TAL).
5. Converted the proportion of significant NUC and TAL findings (for the four measures of aging) into Z-scores based on the null distributions

1. Generated 100 random gene / data sets (~320 genes each for NUC & 16 genes each for TAL) of the eQTL data.
2. Ran MR for the random sets for each of the aging phenotypes for NUC and TAL.
3. Calculated the proportion of significant findings for the null pathways.
4. Generated eight null distributions (4 NUC and 4 TAL).
5. Converted the proportion of significant NUC and TAL findings (for the four measures of aging) into Z-scores based on the null distributions.
6. Can now compare how enriched the NUC and TAL findings are across the measures of aging & with each other.

What influences healthspan?

Fishing for intermediate phenotypes…

1. Metabolites impact aging
   • Did MR screen of ~115 metabolites on the aging phenotypes

- Results sparked a very specific hypothesis

"L.LDL.P", "ApoB", "L.LDL.CE", "L.LDL.C", "L.LDL.FC", "Est.C",    
"L.LDL.PL", "M.LDL.PL", "L.LDL.L"

1. NUC genes impact LDL cholesterol
   • Tested this hypothesis!

• HEIDI-verified top finding (will describe later) is PELO.
• PELO is involved in ribosomal biogenesis: assembly and disassembly.
• Homozygous knockouts are embryonic lethal

MAF 0.03; EA=G; REF=T; decreases expression -1.282; decreases LDL -0.05

Antagonistic pleiotropy, where not enough kills you early in life. But later in life, 97% of the population have the other allele. Thus, perhaps a population-level predisposition to higher LDL!

Examples:

• Chicken project
  • How to obtain SRA-stored fastq files
  • Knowing whether the data are pair-end reads…
• RNAseq files in a matrix (easy to get, but…)
  • Figuring out how the authors named their columns
  • Figuring out how the gene names were reported & fetching gene symbols

Aging phenotyes

1.  Timmers, P. R. et al. Genomics of 1 million parent lifespans 
implicates novel pathways and common diseases and distinguishes 
survival chances. eLife 15, (2019).

--parental lifespan GWAS (n=1,012,240), coded as a protective ratio,
negation of the hazard ratio; probability living longer

2.  Zenin, A. et al. Identification of 12 genetic loci associated 
with human healthspan. Commun Biol 2, (2019).

--healthspan (n=300,447), disease-free living 
(time to death, heart disease, dementia, etc...); coded as a hazards ratio; 
p-value reported as -log10(p-value)

Aging and eqtl GWAS

3.  Deelen, J. et al. A meta-analysis of genome-wide association 
studies identifies multiple longevity genes. Nat Commun 10, (2019).

--longevity GWASs for 90/99th vs 60th (n=36,745/28,967); no RSIDs

4. Võsa, U. et al. Unraveling the polygenic architecture of 
complex traits using blood eQTL meta-analysis. bioRxiv (2018)
doi:10.1101/447367.

--expression (eQTL) GWASs (n=30,596), results reported as 1) binary files! 
2) text files but with Z-scores (no betas or standard errors)

Metabolites and LDL GWASs

5.  Kettunen, J. et al. Genome-wide study for circulating 
metabolites identifies 62 loci and reveals novel systemic 
effects of LPA. Nat Commun 7, (2016).

--GWASs of (NMR-quantified) circulating metabolites: n=24,925
123 metabolites (~115 useful for MR), lipids, fatty acids, &
amino acids (using as instrument)

6.  Collins, R. What makes UK Biobank special? Lancet 31, 
1173–1174 (2012).

--UK Biobank for large GWAS of LDL: n=440,546 (using as outcome) 

1.  Davey Smith, G. D. & Ebrahim, S. “Mendelian randomization”: 
can genetic epidemiology contribute to understanding 
environmental determinants of disease? Int J Epidemiol vol. 32 (2003).
-- Misleading to think MR is focused on "results"

2.  Zhu, Z. et al. Integration of summary data from GWAS 
and eQTL studies predicts complex trait gene targets. 
Nat Genet 48, 481–487 (2016).
-- HEIDI, post-results analysis: null of pleoitropy vs linkage; 
contacted the lead author in Australia; taught myself 

3.  Bulik-Sullivan, B. et al. An atlas of genetic correlations 
across human diseases and traits. Nat Genet 47, 1236–1241 (2015).
-- LD Score Regression; genetic correlations, analygous to 
polygeneic-risk scores; taught myself

• Pleiotropy
  • genes controlling more than one trait
  • MR is based on pleiotropy
  • MR also cannot completely rule out unwanted pleiotropy…
• Linkage disequilibrium (LD)
  • causally used synonymously with “linkage”
  • means “genetically associated”

Model of causality

Although I’ve given a taste of some findings, I’m at the stage of analyzing the results & performing the post-results analyses…

A lot more work to do to sift through the MR and HEIDI findings.

• NUC pathway enriched for healthspan
• TAL is good positive control
• Four phenotypes of aging highly genetically correlated
• Overlap in individual NUC genes impacting aging phenotypes
• We can prioritize those that are minimally pleiotropic for functional follow-up
• Some NUC genes impacts on LDL