sharp

A single-cell demultiplexing pipeline.

the sharp (♯) from musical notation similar to the hash (#) in hashtag.

Sharp supports the following:

• Hashtag
• CITE-seq
• ASAP-seq
• CellPlex

Outputs

In addition to cell barcode/UMI correction and quantification, cells are demultiplexed to their sample-of-origin and cross-sample doublets are identified (except CITE-seq). More information about the outputs can be found here

Setup

Sharp is a part of SCING (Single-Cell pIpeliNe Garden; pronounced as "sing" /siŋ/). For setup, please refer to this page. All the instructions below is given under the assumption that you have already configured SCING in your environment.

Create Job Files

You need two files for processing a sample - one inputs file and one labels file. Use the following example files to help you create your job file:

configs/
├── asapseq-tsa.inputs.json
├── asapseq-tsa.labels.json
├── cellplex.inputs.json
├── cellplex.labels.json
├── citeseq.inputs.json
├── citeseq.labels.json
├── hashtag-10x-tsc.inputs.json
├── hashtag-10x-tsc.labels.json
├── hashtag-10x-v3-tsa.inputs.json
├── hashtag-10x-v3-tsa.labels.json
├── hashtag-10x-v3-tsb.inputs.json
├── hashtag-10x-v3-tsb.labels.json
├── hashtag-indrop-methanol.inputs.json
└── hashtag-indrop-methanol.labels.json

Inputs

"Hashtag.resourceSpec": {
    "cpu": 4,
    "memory": -1
},

Setting memory to -1 will tell Sharp to estimate the amount of memory required, but sometimes you might end up needing more. Change it to 256, for example, if you want to tell Sharp to allocate 256 GB memory.

Submit Your Job

Hashtag

conda activate scing

./submit-hashtag.sh \
  -k ~/keys/cromwell-secrets.json \
  -i configs/hashtag-10x-v3-tsb.inputs.json \
  -l configs/hashtag-10x-v3-tsb.labels.json \
  -o Sharp.options.aws.json

CITE-SEQ

conda activate scing

./submit-citeseq.sh \
  -k ~/keys/cromwell-secrets.json \
  -i configs/citeseq.inputs.json \
  -l configs/citeseq.labels.json \
  -o Sharp.options.aws.json

ASAP-seq

conda activate scing

./submit-asapseq.sh \
  -k ~/keys/cromwell-secrets.json \
  -i configs/asapseq-tsa.inputs.json \
  -l configs/asapseq-tsa.labels.json \
  -o Sharp.options.aws.json

CellPlex

conda activate scing

./submit-cellplex.sh \
  -k ~/keys/cromwell-secrets.json \
  -i configs/cellplex.inputs.json \
  -l configs/cellplex.labels.json \
  -o Sharp.options.aws.json

Manual Inspection

Starting from sharp v0.1.0, basic QC notebooks are automatically generated by the pipeline for hashtag and CITE-seq. Below is instructions for manually generating a QC notebook only if necessary, but the QC notebook might be a bit outdated.

Setup

Jupyter Notebook and Papermill are required.

$ conda create -n sharp python=3.8 pip
$ conda activate sharp
$ pip install -r requirements.txt

How to Generate QC Notebook Manually

$ cd manual-inspection
$ ./run.sh
USAGE: run.sh [options]
    -k  service account key (e.g. secrets.json)
    -t  type ('hashtag', 'citeseq', 'asapseq', or 'cellplex')
    -w  workflow ID
    -s  skip download and use the pre-downloaded data

$ ./run.sh -k ~/keys/cromwell-secrets.json -t hashtag -w 47080814-0fe7-458d-9edb-5e3cb86bf870