Slide-seq Pipeline

This is the new alignment pipeline for Slide-seq data. For more information about Slide-seq method itself, see Rodriques & Stickels et al. and Stickels et al..

Quickstart

Install

conda install uses a fair amount of RAM, so you should do this from an interactive session and not on the login node.

use UGER  # to make the ish command available
ish -l h_vmem=4G  # create an interactive session with extra memory
git clone https://github.com/MacoskoLab/slideseq-tools.git  # clone this repository
cd slideseq-tools
conda env create -f environment.yaml  # creates an environment named `slideseq`

You can name the environment something else by adding -n [env_name]. There's no reason to do this, but I figured out how to make it work with an arbitary env name, so I wanted to mention it.

See the bottom of this README for instructions on setting up gcloud credentials.

Submitting a flowcell

Again, you should do this from an interactive session, as downloading the worksheet is a little too strenuous for the login node.

use UGER  # for submitting jobs to the cluster
use Google-Cloud-SDK  # for accessing the Google worksheet
conda activate slideseq
submit_slideseq RUN [RUN...]

This is will submit a set of jobs to process the flowcell(s). You will get emails when the jobs start and end.

One job per lane of the flowcell, for demuxing
For each lane, an array of alignment jobs for each library sequenced
An array of processing jobs, one for each library

If you have already demuxed the flowcell and are just rerunning alignment, you can use the --no-demux flag. If you are just rerunning post-alignment processing, you can use the --no-align flag. Note these options will overwrite existing data in the library folders!

Build a reference

This can be performed from the login node.

use UGER
conda activate slideseq
build_ref --genome-name [GENOME_NAME] --reference-fasta path/to/genome.fasta --reference-gtf path/to/genes.gtf

This will submit a job to build the reference. You will get emails when the jobs start and end.

When the job is complete, the reference can be specified in the Google sheet as /broad/macosko/reference/[GENOME_NAME]/[GENOME_NAME].fasta

Existing references:

Mouse: /broad/macosko/reference/GRCm39.103/GRCm39.103.fasta
Human: /broad/macosko/reference/GRCh38.102/GRCh38.102.fasta

Overview

Goals

Demultiplex sequencing data from Illumina binary files
Match sequenced barcodes to a known set of slideseq puck barcodes to get spatial information
Align reads to a genome using STAR
Count features and summarize as a gene-cell matrix
Do all of this fast, efficiently, and reproducibly. When possible we'd like to disentangle the different steps so that we can perform them separately.

Explicit Non-Goals

Won't work in arbitary environments. This pipeline is built to run at the Broad on the local computing infrastructure. Hopefully it is clear what is happening and the pipeline can be translated, but it is not a priority for us.
Doesn't allow much flexibility in the workflow. This makes the entire process simpler to configure and design.

Currently unsupported or not yet implemented

Plot of read 1 base distribution
UP distance plots
Bead types other than 180402
Multiple alignments of one library (e.g. exonic and also exonic+intronic)
Probably other stuff...

Requirements

The pipeline is designed to work from an Anaconda3 environment on the UGER cluster. We provide an environment.yml file with the requirements.

Other requirements

The following tools are also needed:

Picard (from the Broad)
Drop-seq tools (from the McCarroll lab)
Java-1.8 for the above
Google-Cloud-SDK to access our metadata tracking sheet

If you are running the pipeline on UGER, you do not need to install any of these tools.

The locations of these tools are in the config.yaml file in the slideseq package, along with other configurable paths.

Config tips

Setting up Google-Cloud-SDK

You will need to authenticate the first time you run the pipeline. This just sets up your Google credentials on UGER so that it knows you have access to the worksheet. You shouldn't have to do this again after the first time. If you have any permissions issues you might not have the correct permissions in our account, let us know and we can fix it.

use Google-Cloud-SDK
gcloud init
gcloud auth login
gcloud auth application-default login

# ... follow instructions

Modifying `.my.bashrc`

You can add the following to .my.bashrc to avoid typing them each time:

use UGER
use Google-Cloud-SDK

This might make login a tiny bit slower, but saves typing and errors from forgetting.