AlexsLemonade / scRNA-seq_sandbox

Exploring some scRNA-seq methodologies, not a full-fledged pipeline ready for use

scRNA-seq_workflow

About the scripts in this repository:

1) Pre-processing pipeline:

For starting from scratch with fastq files, an experiment ID or set of SRA's you would like to process into a counts gene matrix file.

A) For Smart-seq2 data or other single cell data that is available on SRA and not droplet or tag-based.

B) For 10X Genomics data (or Drop-seq data).

2) Post-processing pipeline:

For starting from counts gene matrix data file that you would like to normalize and do further analyses on.

Step 0: Set up docker image

For either pipeline, first clone the repository and then set up the docker image to work from: Assuming you have docker installed already on your computer. Follow these steps to run this. Open up command line (Terminal or what have you).

1. Build a docker image with the Dockerfile and create a container

$ docker build -< Dockerfile -t <DESIRED_IMAGE_TAG_HERE>

2. Run a container with this image

$ docker run -it --rm --mount type=volume,dst=/home/rstudio/kitematic,volume-driver=local,volume-opt=type=none,volume-opt=o=bind,volume-opt=device=<PUT_DESIRED_LOCAL_DIRECTORY_PATH_HERE> -e PASSWORD=<DESIRED_PASSWORD_HERE> -p 8787:8787 <SAME_DESIRED_IMAGE_TAG_AS_ABOVE_HERE>

3. Go to the command line in your container:

Run the first line so you can find out what your container id is.

$ docker ps

It will be something like "a1b23c45" (a jumble of lower case letters and numbers). And you'll put that here:

$ docker exec -it <CONAINER_ID> bash

1) Pre-processing pipeline:

For starting from scratch with an experiment ID or set of SRA's you would like to process into a counts gene matrix file.

A) For Smart-seq2 data or other single cell data that is available on SRA and not droplet or tag-based.

Step A1. Open run_pre-processing_pipeline.sh and change the variables to the dataset you

are working with OR keep this the same and follow this example's dataset.

# Change your directory name, GEO ID, and SRP here. Then run the script.
dir=darmanis_data
GSE=GSE84465
SRP=SRP079058
label=darmanis

Step A2. Run the pipeline to process the data with salmon and tximport to create a gene matrix file

Depending on how many samples are in the dataset this will take an hour or days (if you have thousands of samples)

Step A3. Open up Rstudio and prep the data for post-processing

To open Rstudio in docker, go to your internet browser and enter: localhost:8787 Follow the example in darmanis_data_prep.Rmd to set up data.

$ cd <PATH_TO_THE_CLONED_REPOSITORY>  
$ bash run_pre-processing_pipeline.sh

B) For 10X Genomics data (or Drop-seq data)

Step B1. Open run_tag-based_pre-processing_pipeline.sh and change the url and variables to the dataset you

are working with OR keep this the same and follow this example's dataset.

# Change your directory name, and label here.
dir=pbmc_data
label=pbmc_1k_v2

Change this line to the url of the fastq files for the dataset you want to work with. Or keep as is and follow the example

cd ${dir}
curl -O http://cf.10xgenomics.com/samples/cell-exp/3.0.0/pbmc_1k_v2/pbmc_1k_v2_fastqs.tar
tar -xvf ${label}.tar

Step B2. Run the pipeline to process the data with Alevin to create a gene matrix file

Depending on how many samples are in the dataset this will take a few hours or so.

2) Post-processing pipeline:

For starting from counts gene matrix data file that you would like to normalize and do further analyses on.

Step 1. Open up Rstudio and prep the data for post-processing

To open Rstudio in docker, go to your internet browser and enter: localhost:8787 Follow the example in darmanis_data_prep.Rmd to set up data.

Step 2. Open run_post-processing_pipeline.sh and change the variables to the dataset you

are working with OR keep this the same and follow this example's dataset.

dir=darmanis_data
label=darmanis

Step 3. Run the pipeline

$ cd <PATH_TO_THE_CLONED_REPOSITORY>  
$ bash run_post-processing_pipeline.sh