QTL and GWAS

class: center, middle, inverse, title-slide

.title[
# QTL and GWAS
]
.author[
### Jinliang Yang
]
.date[
### May 3, 2024
]

---

# QTL or Linkage Mapping

Linkage mapping uses statistical techniques to localize chromosomal
regions that might contain genes contributing to phenotypic variation in a complex trait of interest.

#### Steps for QTL mapping include:

- 1) Create a segregating population
    - __mating design__
  
  - 2) Genotype individuals within this population with molecular markers
    - __genotyping by sequencing__
  
  - 3) Phenotype the individuals
    - __field experimental design__
  
  - 4) Apply statistical models to associate the markers to the phenotypic variation
    - __single-marker analysis__ or __interval mapping__

---
# Recombinant inbred lines (RILs)

.pull-left[
<div align="center">
<img src="ril.jpg" height=400>
</div>
> Mauricio et al., 2001
]

#### a. Select the parental lines with extreme phenotypes

- The density of hairs that occur on a plant leaf

#### b. Selfing an F1 to form a population of F2 individuals

#### c. Each F2 is selfed for six additional generations

- Forming a population of recombinant inbred lines (RILs)

- Each RIL is homozygous for a section of a parental chromosome

---
# Simulating an F2 population

```r
library(qtl)
set.seed(12347)
# Five autosomes of cM length 50, 75, 100, 125, 60
L <- c(50, 75, 100, 125, 60)
map <- sim.map(L, n.mar=L/5+1, eq.spacing=FALSE, include.x=FALSE)
# Simulate a F2 population with 250 individuals
pop <- sim.cross(map, type="f2", n.ind=250, model = rbind(c(1,45,1,1),c(5,20,0.5,-0.5)))
plot.map(pop)
```

---

# Simulating a QTL mapping experiment

#### Checking phenotypic distribution

```r
hist(pop$pheno$phenotype, main="simulated phenotype", 
     breaks=50, xlab="Phenotype", col="#cdb79e")
```

---
# Single-marker analysis

```r
# single-QTL scan by marker regression with the simulated data
out.mr <- scanone(pop, method="mr")

# plot of marker regression results for chr 4 and 12
plot(out.mr, chr=c(1,2,3,4,5), ylab="LOD Score")
```

---
# Haley-knott Regression

This is a version of interval mapping which is a very good approximation to interval mapping via maximum likelihood.

```r
# single-QTL scan using Haley-knott Regression approach
out.hk <- scanone(pop, method="hk")

# plot of marker regression results for chr 4 and 12
plot(out.hk, chr=c(1,2,3,4,5), ylab="LOD score")
```

---
# Plot QTL effect

```r
# summary of out.mr
summary(out.mr, threshold=3)
```

```
##       chr  pos  lod
## D1M10   1 42.3 21.7
## D5M6    5 21.0  4.8
```

```r
effectplot(pop, mname1="D1M10", main="Chr1")
```

---
# Plot QTL effect

```r
# summary of out.mr
summary(out.mr, threshold=3)
```

```
##       chr  pos  lod
## D1M10   1 42.3 21.7
## D5M6    5 21.0  4.8
```

```r
effectplot(pop, mname1="D5M6", main="Chr5")
```

---

# QTL recap

.pull-left[
<div align="center">
<img src="ril2.png" height=400>
</div>
> Candela and Hake, 2008
]

.pull-right[
### Traditional bi-parental QTL:

- Newly created recombination events (very few)

- Low mapping resolution (Mb level, ~ hundreds of genes)

- Limited number of QTLs

- Can be used for marker assistant selection (MAS)
]

---

# GWAS: A brief overview

.pull-left[
<div align="center">
<img src="gwas1.png" height=400>
</div>

]

.pull-right[
### GWAS in a nutshell:

- Utilize existing populations

- Leverage historical recombination events

- Achieve high mapping resolution but need high density marker
]

---

# Nested association mapping (NAM)

<div align="center">
<img src="nam.png" height=450>
</div>
> Yu et al., 2008

---

# SNP projection for NAM RILs

- More than 15 years ago

- No reference genome available and the genotyping is challenging

- It is a cost effective method to identify trait-associated genes in high resolution

---

# A diversity panel for GWAS

.pull-left[
<div align="center">
<img src="diversity.png" height=300>
</div>
> Liu et al., 2003

- NSS, SS, TS, popcorn, sweet corn, mixed group
]

.pull-right[
<div align="center">
<img src="founders.png" height=190>
</div>

- With diverse genetic backgrounds repsenting different subpopulations

- Individuals with known or unknown predigrees or ancestral origins

- Representative sampling of alleles across the genome to capture genetic variation
]

---

# Population structure and  relatedness

__Subpopulations (Q) __: ancestral origins and subgroups due to population subdivision

__Genetic relatedness (K)__: the degree of genetic similarity within a subpopulation

---

# Mixed linear model for GWAS

<div align="center">
<img src="mlm.png" height=330>
</div>
> Yu et al., 2006

- __False Discovery Rate (FDR)__: the proportion of __false positive__ results among all statistical significant associations identified in a GWAS

- __GWAS Power__: the ability of a GWAS to detect __true positive__ genetic associations with phenotypic traits

---

# Other strategies for association analysis

.pull-left[

### Extreme Phenotype GWAS (or XP-GWAS)

<div align="center">
<img src="krn.png" height=330>
</div>
> J. Yang et al., 2015
]

.pull-right[
#### Pros:
- Increased power due to variants likely have large effect size

- Cost effective for genotyping and no need for additional phenotyping

#### Cons:
- Limited generalizability: only for a given trait

- Collecting large sample size can be challenging

- Extreme phenotype selection may introduce bias if differs systematically
]

---

# Other strategies for association analysis

.pull-left[

### Mediation Annalysis

<div align="center">
<img src="med.png" height=200>
</div>
> Z. Yang et al., 2022

__dSNP__: directly associated SNP
__iSNP__: indirectly associated SNP
__mediator__: gene mediating via iSNP
]

.pull-right[
#### Pros:
- Establish a assumed causal chain from SNP to gene expression to phenotype

- Mapping down to the gene (mediator)

#### Cons:
- Population-wide Omics data is required

- Currently only powerful for large effect mediators
]

---
# Multi-Omics data integration