1. Computational personal genomics: selection, regulation, epigenomics, disease
Manolis Kellis
Broad Institute of MIT and Harvard
MIT Computer Science & Artificial Intelligence Laboratory

2. Understanding human variation and human disease
Gene annotation
(Coding, 5’/3’UTR, RNAs)
Evolutionary signatures
Roles in gene/chromatin regulation
 Activator/repressor signatures
CATGACTG
CATGCCTG
Disease-associated
variant (SNP/CNV/…)
Non-coding annotation
 Chromatin signatures
Other evidence of function
 Signatures of selection (sp/pop)
Challenge: from loci to mechanism, pathways, drug targets
Goal: A systems-level understanding of genomes and gene regulation:
The regulators: Transcription factors, microRNAs, sequence specificities
The regions: enhancers, promoters, and their tissue-specificity
The targets: TFstargets, regulatorsenhancers, enhancersgenes
The grammars: Interplay of multiple TFs  prediction of gene expression
 The parts list = Building blocks of gene regulatory networks
add cartoon image here (remember slide is copied below) 2

3. Compare 29 mammals: Reveal constrained positions
NRSF motif
Reveal individual transcription factor binding sites
Within motif instances reveal position-specific bias
More species: motif consensus directly revealed

4. Chromatin state dynamics across nine cell types
Predicted
linking
Key points to make:
Chromatin states enabled us to study the dynamic nature of chromatin across many cell types.
By distinguishing 15 different types of chromatin states, we could summarize all significant combinations of 81 different chromatin tracks and 2.4 billion reads in just nine chromatin annotation tracks, one for each cell type.
For example, the same gene (WLS), is ‘poised’ in embryonic stem cells (ES), repressed in three other cell types (K562, blood, and liver), and active in the other five cell types.
This allows us to now define ‘vectors’ of activity for each region of the genome, based on the chromatin annotation in the nine cell types.
Correlated activity
Single annotation track for each cell type
Summarize cell-type activity at a glance
Can study 9-cell activity pattern across

5. Revisiting disease- associated variantsxx
Revisiting disease- associated variants
Disease-associated SNPs enriched for enhancers in relevant cell types
E.g. lupus SNP in GM enhancer disrupts Ets1 predicted activator

6. HaploReg: Automate search for any disease study (compbio. mit
HaploReg: Automate search for any disease study (compbio.mit.edu/HaploReg)
Start with any list of SNPs or select a GWA study
Mine publically available ENCODE data for significant hits
Hundreds of assays, dozens of cells, conservation, motifs
Report significant overlaps and link to info/browser

7. 54000+ measurements (x2 cells, 2x repl)
Experimental dissection of regulatory motifs for 10,000s of human enhancers
measurements (x2 cells, 2x repl)

8. Example activator: conserved HNF4 motif match
WT expression
specific to HepG2
Motif match disruptions reduce expression to background
Non-disruptive changes maintain expression
Random changes depend on effect to motif match

9. Allele-specific chromatin marks: cis-vs-trans effects
Maternal and paternal GM12878 genomes sequenced
Map reads to phased genome, handle SNPs indels
Correlate activity changes with sequence differences

10. Brain methylation in 750 Alzheimer patients/controls
500,000
methylation
probes
750 individuals
Brad Bernstein
REMC mapping
Phil de Jager, Roadmap disease epigenomics
Genome
Epigenome
meQTL
Phenotype
Classification
MWAS 1 2
10+ years of cognitive evaluations, post-mortem brains
93% of functional epigenomic variation is genotype driven!
Global repression in 7,000 enhancers, brain-specific targets

11. Global hyper-methylation in 1000s of AD-associated loci
Top 7000 probes
P-value
480,000 probes, ranked by Alzheimer’s association
Methylation
Alzheimer’s-associated probes are hypermethylated
Global effect across 1000s of probes
Rank all probes by Alzheimer’s association
7000 probes increase methylation (repressed)
Enriched in brain-specific enhancers
Near motifs of brain-specific regulators
Complex disease: genome-wide effects

12. Covers computational challenges associated with personal genomics:
- genotype phasing and haplotype reconstruction  resolve mom/dad chromosomes
- exploiting linkage for variant imputation  co-inheritance patterns in human population
- ancestry painting for admixed genomes  result of human migration patterns
- predicting likely causal variants using functional genomics  from regions to mechanism
- comparative genomics annotation of coding/non-coding elements  gene regulation
- relating regulatory variation to gene expression or chromatin  quantitative trait loci
- measuring recent evolution and human selection  selective pressure shaped our genome
- using systems/network information to decipher weak contributions  combinatorics
- challenge of complex multi-genic traits: height, diabetes, Alzheimer's  1000s of genes

13. Personal genomics today: 23 and We
Recombination breakpoints
Family Inheritance
Me vs.
my brother
My dad
Dad’s mom
Mom’s dad
Human ancestry
Disease risk
Genomics: Regions  mechanisms  drugs
Systems: genes  combinations  pathways

14. Personal genomics tomorrow: Already 100,000s of complete genomes
Health, disease, quantitative traits:
Genomics regions  disease mechanism, drug targets
Protein-coding  cracking regulatory code, variation
Single genes  systems, gene interactions, pathways
Human ancestry:
Resolve all of human ancestral relationships
Complete history of all migrations, selective events
Resolve common inheritance vs. trait association
What’s missing is the computation
New algorithms, machine learning, dimensionality reduction
Individualized treatment from 1000s genes, genome
Understand missing heritability
Reveal co-evolution between genes/elements
Correct for modulating effects in GWAS

15. Collaborators and Acknowledgements
Chromatin state dynamics
Brad Bernstein, ENCODE consortium
Methylation in Alzheimer’s disease
Phil de Jager, Brad Bernstein, Epigenome Roadmap
Mammalian comparative genomics
Kerstin Lindblad-Toh, Eric Lander, 29 mammals consortium
Massively parallel enhancer reporter assays
Tarjei Mikkelsen, Broad Institute
Funding
NHGRI, NIH, NSF Sloan Foundation

16. MIT Computational Biology group Compbio.mit.edu
Mike Lin
Ben
Holmes
Soheil
Feizi
Angela
Yen
Luke
Ward
Bob
Altshuler
Mukul
Bansal
Chris Bristow
Stefan
Washietl
Pouya
Kheradpour
Matt
Eaton
Manolis
Kellis
Jason
Ernst
Irwin
Jungreis
Rachel
Sealfon
Jessica Wu
Daniel
Marbach
Louisa
DiStefano
Dave
Hendrix
Loyal
Goff
Sushmita
Roy
Stata3
Stata4

17. Human constraint matches region activity
Active regions
Average diversity (heterozygosity)
Aggregate over the genome
Conserved regions:
Non-ENCODE regions show increased diversity
 Loss of constraint in human when biochemically-inactive
Non-conserved regions:
ENCODE-active regions show reduced diversity
 Lineage-specific constraint in biochemically-active regions