Merging multiple marvel objects

[Rahul1711arora]

Hi,

I am trying to run some analysis where I have a seurat object with normalised expression matrix and raw expression matrix that are being used as input in marvel object creation along with splice junction files created with starSOLO.
What I want to do is to create a marvel object that has the information from all splice junction files corresponding to each sample. For instance, in the seurat object, I have 6 samples that are merged. Now, I want to create a single marvel object for this merged data such that I can infer the splicing pattern of these samples and do the same with another group of samples stored in another seurat object.
Is there a way to do that? Also, is there a way to identify which cell type does the splicing information correspond to? Because for gene expression, I have cell type specific seurat objects, and I intend to understand for instance how splicing differs in cell type 1 versus cell type 2 for the same transcript.

If you would need to have some data, I can subset my data and can share the object with you.
Thanks in advance for your help.

Best,
Rahul

[szhaoaf]

Hi, Rahul,

What you concerned is absoultely solvable. You can simply read in your STARsolo SJ matrix as scRNAseq matrix，then perform merging and extracting. Below is the codes I used for multiple dataset analysis ( prepare for SJ related files)

`samples <- c('A','B','C','D','E','F')

for(x in samples){
sj_name = read.table(paste0(x,"_Solo.out/SJ/raw/features.tsv"), sep="\t", header=F, stringsAsFactors=FALSE) %>%
dplyr::mutate(coord.intron=paste(V1, V2, V3, sep=":")) %>%
dplyr::select(coord.intron)
write.table(sj_name, sep="\t", row.names=F, col.names=F, quote=F,
file=paste0(x,"_Solo.out/SJ/raw/features_new.tsv"))
}

sce_sj.list = lapply(samples, function(x){
print(x)
fls = list.files(paste0(x,"_Solo.out/SJ/raw"), full.name=T)
if(length(grep("gz",fls))!=3){
for(fl in fls){ gzip(fl, remove=FALSE) }
}
count = Read10X(paste0(paste0(x,"_Solo.out/SJ/raw")),gene.column=1)
sce_sj = CreateSeuratObject(count,project = x)
})

sce_sj = merge(sce_sj.list[[1]], sce_sj.list[-1])
sce_sj <- JoinLayers(sce_sj)

sj_count_merge = sce_sj[['RNA']]$counts

writeMM(sj_count_merge, file = './work/sj/raw/matrix.mtx')

sj_count_merge_phe = sce_sj@meta.data %>%
tibble::rownames_to_column("cell.id") %>%
dplyr::select(cell.id)

write.table(sj_count_merge_phe, sep="\t", row.names=F,
file="./work/sj/raw/phenoData.txt")

sj_count_merge_feat = data.frame(coord.intron=rownames(sce_sj)) %>%
dplyr::mutate(coord.intron=gsub("-","_", coord.intron))

write.table(sj_count_merge_feat, sep="\t", row.names=F,
file="./work/sj/raw/featureData.txt")`

Hope this may offer you help.

Bests,
Shizheng ZHAO