-
Notifications
You must be signed in to change notification settings - Fork 2
/
evaluate_power.Rmd
166 lines (104 loc) · 4.29 KB
/
evaluate_power.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
---
title: "Power calculations for grant"
subtitle: 'Public release of data and code from Hoffman, et al. (submitted)'
author: "Developed by [Gabriel Hoffman](http://gabrielhoffman.github.io/)"
date: "Run on `r Sys.Date()`"
documentclass: article
output:
html_document:
toc: true
---
# library(pwr)
# ensGene = 'ENSG00000214776'
# beta = 2.67
# mu = mean(vobj$E[ensGene,])
# logFC = log2(mu+2*beta) - log2(mu)
# logFC
# FC = (mu+2*beta) / mu
# log2(FC)
# s = sd(2^vobj$E[ensGene,metadata$Cell.Type == levels(metadata$Cell.Type)[2]])
# d = abs(FC / s)
# n1 = 4
# n2 = 12
# pwr.t2n.test(n1=n1, n2=n2, d=d, sig.level=0.05, alternative="greater")
# s2 = sd(vobj$E[ensGene,metadata$Cell.Type == levels(metadata$Cell.Type)[2]])
# pwr.t2n.test(n1=n1, n2=n2, d=abs(logFC/s2), sig.level=0.05, alternative="greater")
<!---
# run analysis
cd /media/sdb1/workspace/scripts/Brennand/COS/COS_public_release
rmarkdown::render("evaluate_power.Rmd")
--->
```{r load.library}
library(snpStats)
library(pwr)
library(foreach)
library(readr)
library(edgeR)
library(data.table)
library(synapseClient)
synapseLogin()
```
```{r download}
geneInfo = read.csv(getFileLocation(synGet( 'syn7113731' )), header=TRUE, stringsAsFactors=FALSE)
# read data
geneCountsIn = read.table(synGet('syn8413221')@filePath, header=TRUE, sep='\t', check.names = FALSE)
# extract gene counts
geneCounts = geneCountsIn[,-c(1:6)]
rownames(geneCounts) = geneCountsIn$Geneid
# identify genes with sufficient expression
isexpr = rowSums(cpm(geneCounts)>1) >= 5
# normalize counts by total library size
dgeObj = DGEList(geneCounts[isexpr,])
dgeObj = calcNormFactors( dgeObj, method="none" )
# convert to log2 RPKM by normalizing by gene legnth
geneExpr_log2RPKM = rpkm(dgeObj, gene.length=geneCountsIn$Length[isexpr], log=TRUE)
expr_CPM = read.table(synGet('syn7253892')@filePath, header=TRUE, sep='\t', check.names = FALSE)
rownames(expr_CPM) = expr_CPM$ensembl_gene_id
expr_CPM = expr_CPM[,-c(1:2)]
colnames(expr_CPM) = gsub("RNA_PFC_", "", colnames(expr_CPM))
colnames(expr_CPM) = gsub("RNA_BP_PFC", "BP", colnames(expr_CPM))
design = read.table(synGet('syn7253894')@filePath, header=TRUE, sep='\t', check.names = FALSE)
rownames(design) = colnames(expr_CPM)
```
```{r load.plink}
# ensGene = 'ENSG00000214776'
# rsid = 'rs3863335'
# testSet = data.frame(gene=c('FURIN', 'TMED7', 'GORASP2', 'SLCO2A1', 'ENSG00000214776'),
# rsid = c('rs4702', 'rs56165083', 'rs6433224', 'rs16842260', 'rs3863335'), stringsAsFactors=FALSE)
geneList = array(read.table("/sc/orga/projects/psychgen/resources/commonmind/sczGWS_predixcanOrColoc2.genes", stringsAsFactors=FALSE))
maxEQTL = read.table("/sc/orga/projects/psychgen/resources/commonmind/gene_maxEqtl_beta.txt", header=TRUE, stringsAsFactors=FALSE)
testSet = data.frame(gene=geneList)
testSet$ensGene = geneInfo$Geneid[match(testSet$gene, geneInfo$geneName)]
idx = match(testSet$ensGene, maxEQTL$gene)
testSet$rsid[!is.na(idx)] = maxEQTL$snp[idx[!is.na(idx)]]
testSet = testSet[!is.na(testSet$rsid),]
# Load genotype data
# only load the site variant of interest
file = '/hpc/users/hoffmg01/commonmind/data/genotypes/imputation/final/hardcall/CM5-pos-imputed-hardcall95'
BIM = as.data.table(read_tsv( paste0(file, ".bim"), col_names=FALSE))
idx = testSet$rsid %in% BIM$X2
data = read.plink( paste0(file, ".bed"), paste0(file, ".bim"), paste0(file, ".fam"), select.snps=testSet$rsid[idx])
```
```{r eval.power}
idx = which(testSet$rsid %in% colnames([email protected]))
pwrCalc = foreach(i = idx, .combine=rbind ) %do% {
ensGene = testSet$ensGene[i]
rsid = testSet$rsid[i]
expr = t(expr_CPM[ensGene,,drop=FALSE])
df = merge(expr, [email protected][,rsid,drop=FALSE], by="row.names")
df[,rsid] = as.numeric(df[,rsid])
df[,rsid][df[,rsid]==0] = NA
j = match(df$Row.names, rownames(design))
D = cbind(SNP=as.numeric(df[,testSet$rsid[i]]), design[j,])
fit = lm(df[,testSet$ensGene[i]] ~ ., D )
summary(fit)
s = sd(residuals(fit)) / 3
logFC = 2*coef(fit)[2]
n1 = 4
n2 = 12
powerCis = pwr.t2n.test(n1=n1, n2=n2, d=abs(logFC/s), sig.level=0.05, alternative="greater")
powerAll = pwr.t2n.test(n1=n1, n2=n2, d=abs(logFC/s), sig.level=0.05/20000)
data.frame(ensGene = ensGene, rsid = rsid, eqtlp = coef(summary(fit))[2,4], logFC = as.numeric(logFC), s = s,
powerCis=powerCis$power, powerAll= powerAll$power)
}
```