CellRanger vs Seurat

[akitrav1]

Hi,

I am very new to single cell sequencing and I have a couple of questions, especially to people using the Seurat package?

How does CellRanger compare to Seurat? Do they complement each other or are they competitors?

Are there specific parts of the single cell pipeline that Cellranger performs better, considering it is developed by 10x itself?

[mgogol]

Seurat and cellranger

cellranger is run on the raw data and produces data that you can read into R with Seurat for downstream analysis. It also does some processing of the data for instant visualization in the cellranger report, but we don't typically use that much further, because it's nice to have more control over which cells you filter and how you treat the data.

There is a package called Loupe from 10x that you can use to explore the cellranger-generated data, but again, a bit black box-y and less control over parameters, less clear how the data was actually treated.

[SeqGeq]

Alternative Analysis Platforms

Seurat is a beautiful R package for one workflow in analyzing data generated from CellRanger (and other scRNA-Seq pipelines), built by some top tier talent at NYU.

This can be utilized via the Seurat plugin developed by FlowJo within the SeqGeq platform which runs the Seurat R package as well as many other tools (in parallel or stacked together) for very deep data analysis and without the need to code in R.

More information available here:

flowjo.com/solutions/seqgeq
exchange.flowjo.com

Best,
Ian Taylor
SeqGeq Tech Support
[email protected]

[mgogol]

Uh, I disagree that Seurat is for "one workflow". It is quite flexible, with excellent documentation for a number of use cases that makes it pretty easy to use, even for people who aren't incredibly experienced with R. And with the script you develop and version information, your analysis is reproducible by other people, which it most likely will not be using a point and click type interface. (Please correct me if I'm wrong).

Feel free to advertise your commercial product, but I feel you were slightly misleading about Seurat's capabilities and thought I would chime in with my own opinion. Have a good day.

[SeqGeq]

I'm a huge fan of the Seurat analysis pipeline (we implemented that as a free plugin for SeqGeq for exactly that reason); I didn't mean to mislead.

To your point about reproducibility, I believe there is some misunderstanding - SeqGeq's saving mechanism creates an archive file (a GeqZip) which allows any researcher to recapitulate and reproduce exactly the analysis created there.

[akitrav1]

Can you analyze data from 10x genomics using SeqGeq? How does Cell Ranger compare to that? Also isn't SeqGeq meant for Illlumina-Biorad product?

[SeqGeq]

Yes SeqGeq can analyze data from 10x's pipeline, after processing in CellRanger.

CellRanger is not a tertiary analysis platform, it produces a data matrix of reads per gene per cell from FASTQ files. This is what SeqGeq takes as an input.

Like FlowJo for flow cytometry data analysis, our team has tried to make SeqGeq "platform agnostic", it can analyze data from every single-cell sequencing pipeline I've run across with ease, and also most bulk RNA seq data.

File formats currently supported include: CSV, TSV, TXT, MTX, HDF5 and ST. If you run into a problem analyzing a particular type of data or accomplishing a desired workflow I would encourage you to reach out for training and troubleshooting. We're here to help.

Best,
Ian Taylor
SeqGeq Tech Support
[email protected]

[mgogol]

I'm a huge fan of the Seurat analysis pipeline (we implemented that as a free plugin for SeqGeq for exactly that reason); I didn't mean to mislead.

To your point about reproducibility, I believe there is some misunderstanding - SeqGeq's saving mechanism creates an archive file (a GeqZip) which allows any researcher to recapitulate and reproduce exactly the analysis created there.

Great, I am glad you are making an effort to be reproducible. I am clearly jumpy about this.

[akitrav1]

Data from 10x chromium does not give transcript isoform level data but at a gene level data? Is it correct? Also, between Illumina-ddseq and 10x, what is the general consensus? Which one is better from both technology and software point of view?

[mgogol]

The cellranger pipeline generates gene level data, but you always have the fastq files if you want to go back and look at them another way. I'm not familiar with illumina ddseq.

[[email protected]]

Single Cell Analysis Alternative

Another alternative is Partek Flow. You can work with data from any platform, including the output from CellRanger (gene/feature barcode matrices, hdf5 files, or even FASTQ files) and perform QA/QC, filtering, normalization, clustering, visualization, classification, statistical analysis, pathway analysis and so on. It's all point & click and you get tech support, so it's super easy to use. Here is more information about our single cell tools. Hope this helps.

[fantasticAI]

I wonder if I got 20 samples of scRNAseq data, and 4 of them are normal, 8 of them are treat1 and 8 of them are treat2. Do you think it would be a good idea to use Seurat to merge them together or use Cellranger aggr ?