CS 374: Relating the Genetic Code to Gene Expression Sandeep Chinchali.

CS 374: Relating the Genetic Code to Gene Expression

Sandeep Chinchali

Outline

1. Basic Gene Regulation2. Gene Regulation and Human Disease3. Measurement Technologies4. Papers5. Future Trends

1. BASIC GENE REGULATION

Human Genome

• 3 billion bases – 2% coding, 5-10% regulatory• Organism’s complexity NOT correlated with

number of genes!– Human (20-25k genes) vs. Rice (51k genes)

• 1 million Regulatory elements enable:– Precise control for turning genes on/off– Diverse cell types (lung, heart, skin)

Regulatory Elements

• ~ 20-25k genes– Expression Modulated by ~ 1 Million cis-reg

elements– Enhancer, Promoters, Silencers

Controlling Gene Expression

• Transcription factors (TFs):– Proteins that recognize sequence motifs in

enhancers, promoters– Combinatorial switches that turn genes on/off

Modulating Gene Expression

Expression Quantitative Trait Locus (eQTL):– Regions where different genotypes correlate with

changes in gene expression

Chromatin Remodelling

http://www.cropscience.org.au/icsc2004/symposia/3/1/1957_dennise-5.gif

2. GENE REGULATION AND DISEASE

Bejerano Lab

Disease ImplicationsSHH

MUTATIONS

•Brain

•Limb

•Other

Bejerano Lab

Limb Enhancer 1Mb away from Gene

SHHlimb

Bejerano Lab

SHH

Enhancer Deletionlimb

DELETE

•Limb

Bejerano Lab

SHH

Enhancer 1bp Substitutionlimb

MUTATIONS

•Limb

Lettice et al. HMG 2003 12: 1725-35

Genome Wide Assocation Study (GWAS):

80% of GWAS SNPs are noncoding (many are eQTLs)

Bejerano Lab

From eQTL to Disease

TAllele specific binding may alter gene expression

Outline

1. Basic Gene Regulation2. Gene Regulation and Human Disease3. Measurement Technologies4. Papers5. Future Trends

MEASUREMENT TECHNOLOGIES

GTEX

eQTLs: Correlating Genotype with Expression

RNA-seq, Microarray

SNP Array, WGS

Measuring Open Chromatin

http://hmg.oxfordjournals.org

Measuring open chromatin – DNase Seq

Sequence open chromatin – map enhancers, promoters …wikipedia

Statistical Overview

• Given: Genotype + Expression Matrix• Problem: Determine eQTLs • Possible Solutions:– Regress homozygous/het genotypes with

expression• Key Problem: – Of many linked SNPs, what is the causal variant?

Enhancer

Outline

1. Basic Gene Regulation2. Gene Regulation and Human Disease3. Measurement Technologies4. Papers5. Future Trends

PAPER 1: DISSECTING THE REGULATORY ARCHITECTURE OF GENE EXPRESSION QTLS

Overview

• HapMap cells + 1000G genotypes• Bayesian Model– Uncertainty over functional SNP– Prior: Whether SNP hits a functional element

(TFBS, promoter, etc)– Upweight effect of SNPs in functional regions

• Results:– eQTLs often in TFBS, open chromatin, not

specifically overrepresented in TATA box

METHODS

1. Associate SNPs with Gene Expression

2. Functional Annotation

3. Adjust p-value based on annotation

RESULTS

eQTNs are enriched in enhancers, promoters

Inactive

Active Promoter/En

hancer

eQTNs are enriched in enhancers, promoters (2)

What is the distribution of eQTNs in regulatory sites?

eQTNs enriched in TF binding sites

What TF families show the highest eQTN enrichments?

PAPER 2: DNASE1 SENSITIVITY QTLS ARE A MAJOR DETERMINANT OF HUMAN EXPRESSION

VARIATION

Overview

• If an allele is correlated with changes in open chromatin, how often does it actually modulate gene expression?

• dsQTL – DNase sensitive QTL• dsQTL vs eQTL– Functional link between changes in chromatin

accessibility, gene expression

DNase Hypersensitive Region

http://hmg.oxfordjournals.org

dsQTL – genotype correlates with extent of open chromatin

How does a dsQTL look?

RESULTS

In what proximity of gene’s TSS do dsQTLs occur?

Changes in open chromatin associated with gene expression levels

How might a dsQTL be an eQTL?

Mechanisms of dsQTLs

In which conformations are dsQTLs also eQTLs?

CONCLUSION

Future Trends

• Denser genotyping + more expression measurements in variety of cell lines– Better power to detect eQTLs with more people

• eQTLs with small effect sizes that additively disrupt disease pathways– Common disease, common variant hypothesis

• Better annotating + understanding genome enhances selection of causal eQTNs

EXTRA SLIDES

Connections to GWAS

Joe Pickrell,, Joint analysis of functional genomic data and genome-wide

association studies of 18 human traits

Joe Pickrell,, Joint analysis of functional genomic data and genome-wide

association studies of 18 human traits

References

• 30: http://stanfordcehg.wordpress.com/2013/12/06/which-genetic-variants-determine-histone-marks/