Skip to content

Commit

Permalink
Added population to IBS comparison.
Browse files Browse the repository at this point in the history 
Also updated data story to use the dplyr package throughout.
  • Loading branch information
@deflaux
deflaux committed Jan 9, 2015
1 parent 0829bba commit 6283daf
Show file tree
Hide file tree
Showing 5 changed files with 88 additions and 57 deletions.
74 changes: 45 additions & 29 deletions results/ibs/1000Genomes-chr22/analyze-ibs-data.Rmd
View file
Original file line number Diff line number Diff line change
Expand Up @@ -71,36 +71,38 @@ PLINK/SEQ](https://raw.githubusercontent.com/deflaux/codelabs/qc-codelab/R/1000G

## Loading IBS Data in R

`ibsFilename1` contains an N^2 x 3 IBS matrix, where N is the size of the
`ibsDataFlowFilename` contains an N^2 x 3 IBS matrix, where N is the size of the
population and each row represents the IBS score for a pair of individuals.

```{r message=FALSE, comment=NA}
ibsFilename1="1000genomes_chr22_ibs.tsv"
ibsDataFlowFilename="1000genomes_chr22_ibs.tsv"
```

`ibsFilename2` contains an NxN IBS matrix, where N is the size of the population
`ibsPlinkSeqFilename` contains an NxN IBS matrix, where N is the size of the population
and each cell of the matrix contains the IBS score of a pair of individuals.

```{r message=FALSE, comment=NA}
ibsFilename2="1000genomes_phase1_chr22_plinkseq_ibs.tsv"
ibsPlinkSeqFilename="1000genomes_phase1_chr22_plinkseq_ibs.tsv"
```

```{r message=FALSE, comment=NA}
library(reshape2)
require(reshape2)
require(dplyr)
ReadIBSFile <- function(ibsFilename, header=FALSE, rowNames=NULL) {
ibsData <- read.table(file=ibsFilename, header=header,
row.names=rowNames, stringsAsFactors=FALSE)
return (ibsData)
}
ibsData1 <- ReadIBSFile(ibsFilename1)
ibsData2 <- ReadIBSFile(ibsFilename2, header=TRUE, rowNames=1)
ibsDataflowData <- ReadIBSFile(ibsDataFlowFilename)
ibsPlinkSeqData <- ReadIBSFile(ibsPlinkSeqFilename, header=TRUE, rowNames=1)
```

Transform the NxN matrix into a N^2 x 3 matrix, where each row represents the
IBS score for a pair of individuals.

```{r message=FALSE, comment=NA}
ibsData2 <- melt(data.matrix(ibsData2))
ibsPlinkSeqData <- melt(data.matrix(ibsPlinkSeqData))
```

Set the column names of the two sets of IBS data consistently.
Expand All @@ -109,8 +111,8 @@ Set the column names of the two sets of IBS data consistently.
ColumnNames <- function(ibsData) {
colnames(ibsData) <- c("sample1", "sample2", "ibsScore")
}
colnames(ibsData1) <- ColumnNames(ibsData1)
colnames(ibsData2) <- ColumnNames(ibsData2)
colnames(ibsDataflowData) <- ColumnNames(ibsDataflowData)
colnames(ibsPlinkSeqData) <- ColumnNames(ibsPlinkSeqData)
```

Make the IBS matrix symmetric.
Expand All @@ -122,7 +124,7 @@ MakeIBSDataSymmetric <- function(ibsData) {
ibsScore=ibsData$ibsScore)
ibsData <- rbind(ibsData, ibsPairsMirrored)
}
ibsData1 <- MakeIBSDataSymmetric(ibsData1)
ibsDataflowData <- MakeIBSDataSymmetric(ibsDataflowData)
```

## IBS Heat Map
Expand All @@ -132,24 +134,23 @@ Exclude the IBS values for a genome and itself, because those values are always

```{r message=FALSE, comment=NA}
ExcludeDiagonal <- function(ibsData) {
ibsData <- subset(ibsData, ibsData$sample1 != ibsData$sample2)
ibsData <- filter(ibsData, ibsData$sample1 != ibsData$sample2)
return (ibsData)
}
ibsData2Sample <- ExcludeDiagonal(ibsData2)
ibsPlinkSeqDataSample <- ExcludeDiagonal(ibsPlinkSeqData)
```

Extract the IBS matrix for a random sample of the individuals.

```{r message=FALSE, comment=NA}
SampleIBSMatrix <- function(ibsData) {
SampleIBSMatrix <- function(ibsData, sampleSize=50) {
individuals <- unique(ibsData$sample1)
sampleSize <- 50
sample <- sample(individuals, sampleSize)
ibsData <- subset(ibsData, ibsData$sample1 %in% sample)
ibsData <- subset(ibsData, ibsData$sample2 %in% sample)
return (ibsData)
}
ibsData2Sample <- SampleIBSMatrix(ibsData2Sample)
ibsPlinkSeqDataSample <- SampleIBSMatrix(ibsPlinkSeqDataSample)
```

Draw a heat map based on the sampled IBS scores.
Expand All @@ -166,7 +167,7 @@ DrawHeatMap <- function(ibsData) {
x="Sample", y="Sample"))
p
}
DrawHeatMap(ibsData2Sample)
DrawHeatMap(ibsPlinkSeqDataSample)
```

## Comparison with [PLINK/SEQ](https://atgu.mgh.harvard.edu/plinkseq/) IBS
Expand All @@ -176,10 +177,9 @@ PLINK/SEQ](https://raw.githubusercontent.com/deflaux/codelabs/qc-codelab/R/1000G
and report the number of differences.

```{r message=FALSE, comment=NA}
mergedIBS <- merge(ibsData1, ibsData2, by = c(colnames(ibsData1)[1:2]))
diffIBS <- abs(mergedIBS$ibsScore.x - mergedIBS$ibsScore.y) < 1e-6
mergedDiffIBS <- cbind(mergedIBS, diffIBS)
colnames(mergedDiffIBS) <- c(colnames(mergedDiffIBS)[1:4], "almostEqualIBS")
mergedIBS <- inner_join(ibsDataflowData, ibsPlinkSeqData, by = c(colnames(ibsDataflowData)[1:2]))
mergedDiffIBS <- mutate(mergedIBS,
almostEqualIBS = abs(mergedIBS$ibsScore.x - mergedIBS$ibsScore.y) < 1e-6)
nrow(mergedDiffIBS[mergedDiffIBS$almostEqualIBS == FALSE,])
```

Expand All @@ -194,17 +194,33 @@ model <- lm(ibsScore.y ~ ibsScore.x, mergedIBS)
summary(model)
```

Compare these results to the pedigree.
Comparison of IBS Results to Pedigree
===================================================

First we fetch the pedigree information from the 1,000 Genomes project.
```{r message=FALSE, comment=NA}
pedigree <- read.delim("ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/working/20130606_sample_info/20130606_g1k.ped", as.is=TRUE)
summary(pedigree)
require(dplyr)
ibs <- filter(ibsData1, sample1 != sample2)
ibs_fam <- inner_join(ibs, select(pedigree, Individual.ID, Family.ID), by=c("sample1" = "Individual.ID"))
ibs_fams <- inner_join(ibs_fam, select(pedigree, Individual.ID, Family.ID), by=c("sample2" = "Individual.ID"))
```

We add the population and family identifiers to each individual in our pair.
```{r message=FALSE, comment=NA}
ibs_sample1_ped <- inner_join(ExcludeDiagonal(ibsDataflowData),
select(pedigree, Individual.ID, Family.ID, Population),
by=c("sample1" = "Individual.ID"))
ibs_both_ped <- inner_join(ibs_sample1_ped,
select(pedigree, Individual.ID, Family.ID, Population),
by=c("sample2" = "Individual.ID"))
ibs_relationship <- mutate(ibs_both_ped,
relationship=ifelse(Family.ID.x == Family.ID.y,
"Same Family",
ifelse(Population.x == Population.y,
"Same Population",
"None")))
```

And plot the scores for pairs by their relationship type:
```{r ibs-boxplot, fig.align="center", fig.width=10, fig.height=10, message=FALSE, comment=NA}
boxplot(ibsScore~Family.ID.x == Family.ID.y, data=ibs_fams, main="Identity By State Results compared to Pedigree",
xlab="Individuals in Same Family", ylab="IBS Score")
boxplot(ibsScore~relationship, data=ibs_relationship, main="Identity By State Results compared to Pedigree",
xlab="Relationship Type Between the Pair of Individuals", ylab="Identity By State Score")
```
71 changes: 43 additions & 28 deletions results/ibs/1000Genomes-chr22/analyze-ibs-data.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -67,40 +67,42 @@ PLINK/SEQ](https://raw.githubusercontent.com/deflaux/codelabs/qc-codelab/R/1000G

## Loading IBS Data in R

`ibsFilename1` contains an N^2 x 3 IBS matrix, where N is the size of the
`ibsDataFlowFilename` contains an N^2 x 3 IBS matrix, where N is the size of the
population and each row represents the IBS score for a pair of individuals.


```r
ibsFilename1="1000genomes_chr22_ibs.tsv"
ibsDataFlowFilename="1000genomes_chr22_ibs.tsv"
```

`ibsFilename2` contains an NxN IBS matrix, where N is the size of the population
`ibsPlinkSeqFilename` contains an NxN IBS matrix, where N is the size of the population
and each cell of the matrix contains the IBS score of a pair of individuals.


```r
ibsFilename2="1000genomes_phase1_chr22_plinkseq_ibs.tsv"
ibsPlinkSeqFilename="1000genomes_phase1_chr22_plinkseq_ibs.tsv"
```


```r
library(reshape2)
require(reshape2)
require(dplyr)

ReadIBSFile <- function(ibsFilename, header=FALSE, rowNames=NULL) {
ibsData <- read.table(file=ibsFilename, header=header,
row.names=rowNames, stringsAsFactors=FALSE)
return (ibsData)
}
ibsData1 <- ReadIBSFile(ibsFilename1)
ibsData2 <- ReadIBSFile(ibsFilename2, header=TRUE, rowNames=1)
ibsDataflowData <- ReadIBSFile(ibsDataFlowFilename)
ibsPlinkSeqData <- ReadIBSFile(ibsPlinkSeqFilename, header=TRUE, rowNames=1)
```

Transform the NxN matrix into a N^2 x 3 matrix, where each row represents the
IBS score for a pair of individuals.


```r
ibsData2 <- melt(data.matrix(ibsData2))
ibsPlinkSeqData <- melt(data.matrix(ibsPlinkSeqData))
```

Set the column names of the two sets of IBS data consistently.
Expand All @@ -110,8 +112,8 @@ Set the column names of the two sets of IBS data consistently.
ColumnNames <- function(ibsData) {
colnames(ibsData) <- c("sample1", "sample2", "ibsScore")
}
colnames(ibsData1) <- ColumnNames(ibsData1)
colnames(ibsData2) <- ColumnNames(ibsData2)
colnames(ibsDataflowData) <- ColumnNames(ibsDataflowData)
colnames(ibsPlinkSeqData) <- ColumnNames(ibsPlinkSeqData)
```

Make the IBS matrix symmetric.
Expand All @@ -124,7 +126,7 @@ MakeIBSDataSymmetric <- function(ibsData) {
ibsScore=ibsData$ibsScore)
ibsData <- rbind(ibsData, ibsPairsMirrored)
}
ibsData1 <- MakeIBSDataSymmetric(ibsData1)
ibsDataflowData <- MakeIBSDataSymmetric(ibsDataflowData)
```

## IBS Heat Map
Expand All @@ -135,25 +137,24 @@ Exclude the IBS values for a genome and itself, because those values are always

```r
ExcludeDiagonal <- function(ibsData) {
ibsData <- subset(ibsData, ibsData$sample1 != ibsData$sample2)
ibsData <- filter(ibsData, ibsData$sample1 != ibsData$sample2)
return (ibsData)
}
ibsData2Sample <- ExcludeDiagonal(ibsData2)
ibsPlinkSeqDataSample <- ExcludeDiagonal(ibsPlinkSeqData)
```

Extract the IBS matrix for a random sample of the individuals.


```r
SampleIBSMatrix <- function(ibsData) {
SampleIBSMatrix <- function(ibsData, sampleSize=50) {
individuals <- unique(ibsData$sample1)
sampleSize <- 50
sample <- sample(individuals, sampleSize)
ibsData <- subset(ibsData, ibsData$sample1 %in% sample)
ibsData <- subset(ibsData, ibsData$sample2 %in% sample)
return (ibsData)
}
ibsData2Sample <- SampleIBSMatrix(ibsData2Sample)
ibsPlinkSeqDataSample <- SampleIBSMatrix(ibsPlinkSeqDataSample)
```

Draw a heat map based on the sampled IBS scores.
Expand All @@ -171,7 +172,7 @@ DrawHeatMap <- function(ibsData) {
x="Sample", y="Sample"))
p
}
DrawHeatMap(ibsData2Sample)
DrawHeatMap(ibsPlinkSeqDataSample)
```

<img src="figure/ibs-heat-map-1.png" title="plot of chunk ibs-heat-map" alt="plot of chunk ibs-heat-map" style="display: block; margin: auto;" />
Expand All @@ -184,10 +185,9 @@ and report the number of differences.


```r
mergedIBS <- merge(ibsData1, ibsData2, by = c(colnames(ibsData1)[1:2]))
diffIBS <- abs(mergedIBS$ibsScore.x - mergedIBS$ibsScore.y) < 1e-6
mergedDiffIBS <- cbind(mergedIBS, diffIBS)
colnames(mergedDiffIBS) <- c(colnames(mergedDiffIBS)[1:4], "almostEqualIBS")
mergedIBS <- inner_join(ibsDataflowData, ibsPlinkSeqData, by = c(colnames(ibsDataflowData)[1:2]))
mergedDiffIBS <- mutate(mergedIBS,
almostEqualIBS = abs(mergedIBS$ibsScore.x - mergedIBS$ibsScore.y) < 1e-6)
nrow(mergedDiffIBS[mergedDiffIBS$almostEqualIBS == FALSE,])
```

Expand Down Expand Up @@ -231,7 +231,10 @@ Multiple R-squared: 1, Adjusted R-squared: 1
F-statistic: 1.53e+15 on 1 and 1193554 DF, p-value: < 2.2e-16
```

Compare these results to the pedigree.
Comparison of IBS Results to Pedigree
===================================================

First we fetch the pedigree information from the 1,000 Genomes project.

```r
pedigree <- read.delim("ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/working/20130606_sample_info/20130606_g1k.ped", as.is=TRUE)
Expand Down Expand Up @@ -269,16 +272,28 @@ summary(pedigree)
```

We add the population and family identifiers to each individual in our pair.

```r
require(dplyr)
ibs <- filter(ibsData1, sample1 != sample2)
ibs_fam <- inner_join(ibs, select(pedigree, Individual.ID, Family.ID), by=c("sample1" = "Individual.ID"))
ibs_fams <- inner_join(ibs_fam, select(pedigree, Individual.ID, Family.ID), by=c("sample2" = "Individual.ID"))
ibs_sample1_ped <- inner_join(ExcludeDiagonal(ibsDataflowData),
select(pedigree, Individual.ID, Family.ID, Population),
by=c("sample1" = "Individual.ID"))
ibs_both_ped <- inner_join(ibs_sample1_ped,
select(pedigree, Individual.ID, Family.ID, Population),
by=c("sample2" = "Individual.ID"))
ibs_relationship <- mutate(ibs_both_ped,
relationship=ifelse(Family.ID.x == Family.ID.y,
"Same Family",
ifelse(Population.x == Population.y,
"Same Population",
"None")))
```

And plot the scores for pairs by their relationship type:

```r
boxplot(ibsScore~Family.ID.x == Family.ID.y, data=ibs_fams, main="Identity By State Results compared to Pedigree",
xlab="Individuals in Same Family", ylab="IBS Score")
boxplot(ibsScore~relationship, data=ibs_relationship, main="Identity By State Results compared to Pedigree",
xlab="Relationship Type Between the Pair of Individuals", ylab="Identity By State Score")
```

<img src="figure/ibs-boxplot-1.png" title="plot of chunk ibs-boxplot" alt="plot of chunk ibs-boxplot" style="display: block; margin: auto;" />
Binary file modified results/ibs/1000Genomes-chr22/figure/ibs-boxplot-1.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified results/ibs/1000Genomes-chr22/figure/ibs-heat-map-1.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified results/ibs/1000Genomes-chr22/figure/ibs-lm-1.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.

0 comments on commit 6283daf

Please sign in to comment.