Comments (1)
Hi!
Assuming you dont have multiome available (only scATAC), you will need to either convert your gene set to regions (as this is the features you have) or the region matrix to a gene accessibility matrix and then score your gene set in that matrix with AUCell. The latter may be the easiest, you can just aggregate the probability scores of the regions surrounding a space around the TSS of the gene. For example:
# Helper function
# Retrieves the cisTopic object regions that intersect with the target regions (e.g. genes, etc...)
# Return is a list of region names (character) split by the @
# queryRegions: GRanges
intersectToRegionSet <- function(cisTopicObject, targetRegions, splitBy=colnames(targetRegions@elementMetadata)[1], minOverlap=0.4, as.data.frame=FALSE)
{
# To do: Check types
if(!is.numeric(minOverlap) || (minOverlap<0 && minOverlap>=1)) stop("minOverlap should be a number between 0 and 1 (percentage of overlap between the regions).")
if(is.null(targetRegions@elementMetadata) | ncol(targetRegions@elementMetadata)==0) targetRegions@elementMetadata <- DataFrame(name=as.character(targetRegions))
if(!splitBy %in% colnames(targetRegions@elementMetadata)) stop("missing gene id annotation")
if(length(table(lengths(targetRegions@elementMetadata[,splitBy]))) >1 ) stop("")
# cisTopicObject
cisTopicRegions <- [email protected]
#elementMetadata(cisTopicRegions)[["regionNames"]] <- names(cisTopicRegions)
cisTopicRegionsOverlap <- findOverlaps(cisTopicRegions, targetRegions,
minoverlap=1, #maxgap=0L, select="all",
type="any", ignore.strand=TRUE)#, ...)
if(minOverlap>0)
{
# In i-cisTarget, the default is 40% minimum overlap. Both ways: It takes the maximum percentage (of the peak or the ict region)
# To reproduce those results:
overlaps <- pintersect(cisTopicRegions[queryHits(cisTopicRegionsOverlap)], targetRegions[subjectHits(cisTopicRegionsOverlap)])
percentOverlapHuman <- width(overlaps) / width(cisTopicRegions[queryHits(cisTopicRegionsOverlap)])
percentOverlapPeaks <- width(overlaps) / width(targetRegions[subjectHits(cisTopicRegionsOverlap)])
maxOverlap <- apply(cbind(percentOverlapHuman, percentOverlapPeaks), 1, max)
cisTopicRegionsOverlap <- cisTopicRegionsOverlap[maxOverlap > minOverlap]
}
if (as.data.frame == FALSE){
hitsPerGene <- split(unname(as.character(cisTopicRegions[queryHits(cisTopicRegionsOverlap)])), unlist(targetRegions[subjectHits(cisTopicRegionsOverlap)]@elementMetadata[,splitBy]))
regionsPerGene <- lapply(hitsPerGene, unique)
missingGenes <- targetRegions@elementMetadata[,splitBy][which(!targetRegions@elementMetadata[,splitBy] %in% names(regionsPerGene))]
regionsPerGene <- c(missing=list(missingGenes), regionsPerGene)
}
if (as.data.frame == TRUE){
hitsPerGene <- cbind(unname(as.character(cisTopicRegions[queryHits(cisTopicRegionsOverlap)])), unlist(targetRegions[subjectHits(cisTopicRegionsOverlap)]@elementMetadata[,splitBy]))
regionsPerGene <- as.data.frame(hitsPerGene[!duplicated(hitsPerGene),])
}
return(regionsPerGene)
}
# Take regions +-10kbp around TSS and in gene's introns
gene2regionFile <- 'Common_data/mm10-limited-upstream10000-tss-downstream10000-full-transcript.bed'
# This is a dataframe with gene and chr \t upstream_boundary \t downstream_boundary
gene2region <- import.bed(con=gene2regionFile)
geneNameSplit <- strsplit(gene2region@elementMetadata$name, split = "#", fixed = TRUE)
geneNameClean <- sapply(geneNameSplit, function(x) x[[1]])
gene2region@elementMetadata$name <- geneNameClean
# Get regions in cisTopicobject per gene
geneRegions <- list()
chunk <- function(x,n) split(x, factor(sort(rank(x)%%n)))
chunks <- chunk(1:length(gene2region),10)
for (i in 1:length(chunks)){
geneRegions <- c(geneRegions, intersectToRegionSet(cisTopicObject, gene2region[chunks[[i]]], splitBy="name", minOverlap=0.4))
}
# Fix duplicated
duplicated_list <- geneRegions[names(geneRegions)[duplicated(names(geneRegions))]]
geneRegions <- geneRegions[-which(names(geneRegions) %in% names(geneRegions)[duplicated(names(geneRegions))])]
for (gene in unique(names(duplicated_list))){
geneRegions[[gene]] <- duplicated_list[[gene]]
}
# Formatting
missingGenes <- geneRegions[["missing"]]
geneRegionSets <- geneRegions[which(!names(geneRegions) %in% "missing")]
pred.matrix <- predictiveDistribution(cisTopicObject)
gene_act <- t(sapply(geneRegionSets, function(x) apply(pred.matrix[x,, drop=F], 2, sum)))
colnames(gene_act) <- [email protected]
gene_act <- round(gene_act * 1000000)
# AUCell
cells_rankings <- AUCell_buildRankings(gene_act)
genes <- c("gene1", "gene2", "gene3") # Your genes
geneSets <- list(geneSet1=genes)
cells_AUC <- AUCell_calcAUC(geneSets, cells_rankings, aucMaxRank=nrow(cells_rankings)*0.05)
# Your signature enrichment score, you can add it as metadata to the cisTopicObject to plot
from cistopic.
Related Issues (20)
- installation issue with ubuntu 20.04 HOT 3
- possibility to use BAM files from bulk ChIP HOT 1
- createcisTopicObject and genomic coordinates incompatibility error HOT 3
- input from fragments.tsv.gz? HOT 1
- Installation issue HOT 2
- Tutorial Dataset files HOT 1
- tSNE Clustering Thresholds HOT 1
- LDA run with Python and LogLikelihhod HOT 1
- How can I run cisTopic in R 4.0 HOT 2
- Ununsed Arguement error in cisTopic/TcisTarget HOT 1
- cisTopicObject <- selectModel(cisTopicObject) Error in .Call("rs_createGD") : C symbol name "rs_createGD" not in load table HOT 1
- Unused argument error while running topicsRcisTarget
- add cisTopic output to Seurat object? HOT 1
- Running Cistopic HOT 1
- Installation error "Failed to install 'unknown package' from Github: ..." HOT 1
- Loading Multiome data
- sha256sum download failed HOT 1
- Error in metadataFeather(path) : Invalid: Not a feather file HOT 3
- Missing resource cleanup in runWarpLDAModels
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from cistopic.