How do I project known tSNE coordinates onto SeuratObject
1
0
Entering edit mode
14 months ago

I would like to analyze a published scRNAseq dataset by making a new Seurat Object. The authors have published their tSNE.1 and tSNE.2 coordinates in addition to all of their metadata but I cannot find how to create a tSNE reduction using their tSNE coordinates.

I'm assuming that I still do basic pre-processing, runPCA and runTSNE but any information on how to project the provided embeddings would be greatly appreciated.

library(dplyr)
library(Seurat)
library(patchwork)
library(ggplot2)
library(data.table)
library(magrittr)
library(Matrix)

data_dir <- 'path/to/data/'
list.files(data_dir) 
expression_matrix <- Read10X(data.dir = data_dir)
fish_LD = CreateSeuratObject(counts = expression_matrix)

metadata <- read.csv('meta.csv') # dataframe I created from barcodes.tsv.gz
Sample <- metadata[1]
Cell.type<- metadata[2]
tSNE.1 <- metadata[3]
tSNE.2 <- metadata[4]
nGene <- metadata[5]
nUMI <- metadata[6]
Percentage.of.mitochondrial.genes <- metadata[7]
Percentage.of.ribosomal.protein.genes <- metadata[8]

fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(Sample = Sample, 
row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(Cell.type = Cell.type, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(tSNE.1 = tSNE.1, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(tSNE.2 = tSNE.2, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(nGene = nGene, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(nUMI = nUMI, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(Percentage.of.mitochondrial.genes = Percentage.of.mitochondrial.genes, row.names = rownames(fish_LD@meta.data)))
fish_LD <- AddMetaData(object = fish_LD, metadata = data.frame(Percentage.of.mitochondrial.genes = Percentage.of.mitochondrial.genes, row.names = rownames(fish_LD@meta.data)))
scRNAseq project single-cell-RNA-seq seurat embeddings • 1.2k views
ADD COMMENT
1
Entering edit mode
14 months ago

If you mean re-run the t-SNE analysis and get exactly the same result? That could be very hard but you should be able to get something similar just by following the standard Seurat pipeline.

To plot the tSNE just use ggplot or something:

ggplot(fish_LD@meta.data,x=tSNE.1,y=tSNE.2,color=Cell.type) + geom_point()
ADD COMMENT
0
Entering edit mode

Thank you, that was extremely helpful and fully recapitulates the original graph. I suppose what I'm trying to do is use seurat functionality (ie. FeaturePlot) but with these tSNE.1 and tSNE.2 coordinates. Any suggestions on how I might accomplish that?

ADD REPLY
0
Entering edit mode

straightforward: add the gene expression as a column in the metadata and change color=Cell.type to color=yourcolumn

there might be some way to do the feature plot normally but i would have to play around with it. you might be able to set dim.1 and dim.2 to fish_LD@meta.data$tSNE.1 etc...

FeaturePlot(object, features.plot, min.cutoff = NA, max.cutoff = NA,
  dim.1 = 1, dim.2 = 2, cells.use = NULL, pt.size = 1,
ADD REPLY

Login before adding your answer.

Traffic: 2727 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6