jebard

library(Seurat)
## Prep for 10x loupe browser inegration
seurat.object 

### first we need to extract out the seurat cell barcodes and convert them back into Loupe space
### note this is tested on integrated objects in seurat with multi samples
seurat.object$loupe_barcodes <- gsub(names(Idents(seurat.object)),pattern = "-1_",replacement = "-")

## pull the embeddings from UMAP
embeds <- (seurat.object@[email protected])

jebard

#!/usr/bin/python
import pysam
import numpy as np
import deeptools.mapReduce
import csv

import argparse

#Tb927_10_v5.1:3648504-3654301 #Tb927.10.14930
#Tb927_10_v5.1:3648504-3654301 (+)

jebard

### handles the Seurat to phate conversio

### first, grab the input required for phate (here we are using the normalized data stored in Seurat
seurat_data <- as.data.frame(seurat.object@assays$RNA@data)

## reshape for input into PHATE
phate_data_input <- t(seurat_data)

## run phate -- this is a default run, feel free to tune params

jebard

wt.subset <- subset(immune.combined, subset = stim == c("CTRL"))
ko.subset <- subset(immune.combined, subset = stim == c("Trp63-"))

FeaturePlot(immune.combined,features = c("Foxi2","Foxc1"),split.by = "stim")

gg1 <- plot_split_by_feature(wt.subset,ko.subset,c("Foxi2"))
gg2 <- plot_split_by_feature(wt.subset,ko.subset,c("Foxc1"))
ggsave(file="Foxi2_Foxc1.png",plot_grid(gg1,gg2,ncol=1))
plot_split_by_feature<- function(wildType, knockout, feature ) {
  ptSize = .4

jebard

library(velocyto.R)
library(pagoda2)
library(Seurat)

### ASSUMES Seurat object has already been computed -- in this case the object cells.combined is the result of seurat

# IMPORTANT!!! create a loom file using the velocyto command line tools first
ldat <- read.loom.matrices("filtered.loom")

# get the spliced and unspliced ratios from the loom file