How to use .rds files for conos?

[yojetsharma]

Hello,
I am a wet-lab stem cell biologist and recently carried out scRNA-seq for 4 neural stem cell lines. So far, I have analysed the scRNA data using the seurat pipeline and also from this workshop https://nbisweden.github.io/workshop-scRNAseq/exercises
I would like to run the analyses again with conos but I am not able to figure out how to start loading the data?
It would be great if someone could break down the initialising steps from me.
Thanks in advance!

[evanbiederstedt]

Hi @yojetsharma

Thanks for using conos.

I'm hoping that this walkthrough will cover all of your questions: https://github.com/kharchenkolab/conos#conos-walkthrough

We have a few more here as well: https://github.com/kharchenkolab/conos#tutorials

Best, Evan

[evanbiederstedt]

Hi @yojetsharma

The input data for conos needs to be Pagoda2 objects or Seurat objects. https://github.com/kharchenkolab/pagoda2/blob/main/doc/pagoda2.walkthrough.md

Given many users work with 10x files, they'll load the data with the pagoda2 function read10xMatrix().

There several tutorials as to how to load the data via Seurat

[yojetsharma]

I did the following and I received the NULL response. Is this correct?

> library(Matrix)
> library(igraph)
> library(pagoda2)
> library(dplyr)
> library(ggplot2)
> countMatrix <- read10xMatrix("/path/to/sample1/filtered_features_bc_matrix_files/")
> countMatrix1 <- read10xMatrix("/path/to/sample2/filtered_features_bc_matrix_files/")
> panel.preprocessed <- lapply(list(countMatrix,countMatrix1), basicP2proc, n.cores=1, min.cells.per.gene=0, n.odgenes=2e3, get.largevis=FALSE, make.geneknn=FALSE)
typeof(panel.preprocessed)
## [1] "list"
names(panel.preprocessed)
NULL

[evanbiederstedt]

@yojetsharma

I'd recommend you start with Section II of the Pagoda2 tutorial and not use the function basicP2proc()

Start at "II. In-Depth Processing and Analysis" here: https://github.com/kharchenkolab/pagoda2/blob/main/doc/pagoda2.walkthrough.md#ii-in-depth-processing-and-analysis

First check that this function and the resulting variable works as expected:

countMatrix <- read10xMatrix("/path/to/sample1/filtered_features_bc_matrix_files/")
dim(countMatrix)
str(countMatrix)

You should be able to take it from there, and do QC.

When you get to Conos, it's a list of Pagoda2 objects: https://github.com/kharchenkolab/conos/blob/main/doc/walkthrough.md

[yojetsharma]

Thanks!
One last doubt: how can I run multiple samples using conos then? Do I need to individually run pagoda2 for all the samples and then proceed to conos?

[evanbiederstedt]

Do I need to individually run pagoda2 for all the samples and then proceed to conos?

That's right

[yojetsharma]

Thank you!
when I reach to conos, how do I merge all the datasets I have and run conos?

[evanbiederstedt]

It should be a list, i.e. list(). Each of the samples in the R list will have a name.

https://github.com/kharchenkolab/conos/blob/main/doc/walkthrough.md

> panel <- conosPanel::panel
> typeof(panel)
[1] "list"
> names(panel)
[1] "MantonBM1_HiSeq_1" "MantonBM2_HiSeq_1" "MantonCB1_HiSeq_1"
[4] "MantonCB2_HiSeq_1"

[yojetsharma]

Okay. I ran my two samples using pagoda2, saved them as .rds files on my system as stated in the tutorial.
Now, do I have upload my .rds files in a repository using drat? And then run "panel <- conosPanel::panel" line?
I am sorry but I cannot understand how to go ahead.

[evanbiederstedt]

You have two samples which are Pagoda2 objects. These are within R.

You then create a list of these objects.

pagoda2object1 <- Pagoda2$new(yourCounts1)
pagoda2object2 <- Pagoda2$new(yourCounts2)

yourList <- list(mysample1 = pagoda2object1 , mysample2 = pagoda2object2)

That's the list you use for conos.

By panel <- conosPanel::panel, I am showing you what the expected data structure should look like.

You might need to ask help from someone in your lab to walk you through these steps.