Differential expression analysis

Last updated: 2018-12-05

workflowr checks: (Click a bullet for more information)

✖ R Markdown file: uncommitted changes
The R Markdown is untracked by Git. To know which version of the R Markdown file created these results, you’ll want to first commit it to the Git repo. If you’re still working on the analysis, you can ignore this warning. When you’re finished, you can run wflow_publish to commit the R Markdown file and build the HTML.
✔ Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
✔ Seed: set.seed(20181119)

The command set.seed(20181119) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.
✔ Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
✔ Repository version: 275d5d8
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
```
Ignored files:
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    analysis/Quality_metrics_cache/
    Ignored:    analysis/figure/

Untracked files:
    Untracked:  analysis/DE_analysis.Rmd
    Untracked:  code/qq-plot.R
    Untracked:  code/summary_functions.R
    Untracked:  docs/figure/DE_analysis.Rmd/

Unstaged changes:
    Modified:   analysis/Quality_metrics.Rmd
    Modified:   analysis/crop_workflow_Alan.Rmd
    Modified:   analysis/index.Rmd
    Modified:   data/DE_input.Rd
```
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

Load data

source("code/summary_functions.R")
load("data/DE_input.Rd")
glocus <- "VPS45"
dim(dm)[1]

NULL

gcount <- dm[1:(dim(dm)[1]-76), colnames(dm1dfagg)[dm1dfagg[glocus,] >0 & nlocus==1]]
# negative control cells defined as neg gRNA targeted cells
ncount <- dm[1:(dim(dm)[1]-76), colnames(dm1dfagg)[dm1dfagg["neg",] >0 & nlocus==1]]
coldata <- data.frame(row.names = c(colnames(gcount),colnames(ncount)),     
                      condition=c(rep('G',dim(gcount)[2]),rep('N',dim(ncount)[2])))

DEseq2

standard DESeq2

library(DESeq2)
dds = DESeqDataSetFromMatrix(countData = cbind(gcount,ncount),
                             colData = coldata,
                             design = ~condition)

dds = estimateSizeFactors(dds)
ddsWARD = DESeq(dds)
resWARD = results(ddsWARD)
summ_pvalues(resWARD$pvalue[!is.na(resWARD$pvalue)])

resSigWARD <- subset(resWARD, padj < 0.1)

There are 0 genes passed FDR <0.1 cutoff.

DESeq2 with LRT test

Following recommendation for single cell from here.

ddsLRT = DESeq(dds, test="LRT", reduced = ~1,  sfType="poscounts", useT=TRUE, minmu=1e-6,minReplicatesForReplace=Inf)
resLRT = results(ddsLRT)
summ_pvalues(resLRT$pvalue[!is.na(resLRT$pvalue)])

resSigLRT <- subset(resLRT, padj < 0.1)

There are 0 genes passed FDR <0.1 cutoff.

edgeR

quasi-likelihood F-tests

library(edgeR)
y <- DGEList(counts= cbind(gcount,ncount),group=coldata$condition)
y <- calcNormFactors(y)
group=coldata$condition
design <- model.matrix(~group)
y <- estimateDisp(y,design)
fitqlf <- glmQLFit(y,design)
qlf <- glmQLFTest(fitqlf,coef=2)
summ_pvalues(qlf$table$PValue)

topTags(qlf)

Coefficient:  groupN 
                       logFC   logCPM        F       PValue          FDR
ENSG00000176956.12 -2.776350 6.601479 58.72481 3.918884e-13 1.169630e-08
ENSG00000100097.11 -2.299929 6.624158 45.44607 1.064435e-10 1.171374e-06
ENSG00000130203.9  -1.831334 6.391155 45.21290 1.177418e-10 1.171374e-06
ENSG00000100300.17 -1.607537 6.410639 40.80264 8.073546e-10 6.024077e-06
ENSG00000138136.6  -1.976572 6.423656 39.44806 1.468212e-09 8.764051e-06
ENSG00000089116.3  -1.542926 6.320288 36.99506 1.605330e-08 7.578892e-05
ENSG00000175899.14 -1.584250 6.857262 33.87858 1.777533e-08 7.578892e-05
ENSG00000162992.3  -1.608906 6.373881 34.49478 2.385458e-08 8.899547e-05
ENSG00000198417.6  -1.600763 6.403261 32.30436 3.635396e-08 1.205578e-04
ENSG00000104327.7  -1.374926 6.326720 49.50835 7.206623e-08 2.150889e-04

likelihood ratio tests

fitlrt <- glmFit(y,design)
lrt <- glmLRT(fitlrt,coef=2)
topTags(lrt)

Coefficient:  groupN 
                        logFC   logCPM       LR       PValue       FDR
ENSG00000175899.14 -1.4640348 6.857262 19.83212 8.454987e-06 0.1170441
ENSG00000176956.12 -2.6064930 6.601479 19.25383 1.144407e-05 0.1170441
ENSG00000100097.11 -2.2150044 6.624158 19.20106 1.176480e-05 0.1170441
ENSG00000100300.17 -1.4959097 6.410639 14.35365 1.514859e-04 0.9651347
ENSG00000119906.11  1.0518706 6.444915 13.90103 1.926926e-04 0.9651347
ENSG00000185900.9  -0.7876253 6.269218 13.68055 2.166870e-04 0.9651347
ENSG00000219626.8  -0.9907215 6.458688 13.59855 2.263601e-04 0.9651347
ENSG00000078061.12 -0.8304381 6.581079 12.89330 3.297606e-04 1.0000000
ENSG00000234912.11  1.0422011 6.412729 12.49191 4.087177e-04 1.0000000
ENSG00000175806.14 -0.7904605 6.612756 12.19362 4.795325e-04 1.0000000

summ_pvalues(lrt$table$PValue)

quasi-likelihood F-tests with prefiltering

Filter genes with 0 coverage in all cells.

mycount <- cbind(gcount,ncount)
totalcount <- apply(mycount,1,sum)
y <- DGEList(counts= mycount[totalcount>0,],group=coldata$condition)
y <- calcNormFactors(y)
group=coldata$condition
design <- model.matrix(~group)
y <- estimateDisp(y,design)
fitqlf <- glmQLFit(y,design)
qlf <- glmQLFTest(fitqlf,coef=2)
topTags(qlf)

Coefficient:  groupN 
                       logFC   logCPM        F       PValue          FDR
ENSG00000176956.12 -2.776350 6.601479 58.72481 3.918884e-13 6.170676e-09
ENSG00000100097.11 -2.299929 6.624158 45.44607 1.064435e-10 6.179874e-07
ENSG00000130203.9  -1.831334 6.391155 45.21290 1.177418e-10 6.179874e-07
ENSG00000100300.17 -1.607537 6.410639 40.80264 8.073546e-10 3.178152e-06
ENSG00000138136.6  -1.976572 6.423656 39.44806 1.468212e-09 4.623693e-06
ENSG00000089116.3  -1.542926 6.320288 36.99506 1.605330e-08 3.998433e-05
ENSG00000175899.14 -1.584250 6.857262 33.87858 1.777533e-08 3.998433e-05
ENSG00000162992.3  -1.608906 6.373881 34.49478 2.385458e-08 4.695177e-05
ENSG00000198417.6  -1.600763 6.403261 32.30436 3.635396e-08 6.360327e-05
ENSG00000104327.7  -1.374926 6.326720 49.50835 7.206623e-08 1.134755e-04

summ_pvalues(qlf$table$PValue)

likelihood ratio tests with prefiltering