WDL for analyzing Drop-seq data
Drop-seq questions may be directed to [email protected].
You may also use this address to be added to the Drop-seq Google group.
See the Drop-seq GitHub repository for more information on the tools.
-
These WDLs are considered early but usable previews, subject to extensive changes!
-
WDLs are separated by into pipelines and tasks
- Pipelines are the top-level WDLs that define the executable workflows.
- Some pipelines are further divided into sub-workflows.
- Tasks are the individual steps that are called by the pipelines that execute a single tool.
- There are experimental "combined" WDLs that aggregate multiple pipelines.
-
The WDLs are designed to
- Run on Terra.bio platform and its underlying Cromwell workflow management system.
- Written in WDL version 1.0.
- Optimize costs on Google Cloud Platform but may run on other platforms.
- Compliment and extend pipelines from
- Optimize Google Compute memory while leaving a buffer, since as of November 2024 Terra.bio does not retry any host VM memory failures. Java Virtual Machines are sized to use a fraction of the host VM memory.
-
Tasks contain a mix of Drop-seq tools and other tools
- Drop-seq tools are from the Drop-seq GitHub repository.
- Other tools are from the Broad Institute GATK and other sources.
- Tasks may expose all or only some of the tool parameters.
| Workflow | Summary |
|---|---|
optimus_post_processing |
Converts Optimus outputs to Drop-seq inputs |
dropseq_cbrb |
Estimates parameters then invokes CBRB |
cell_selection |
Sub-selects barcodes that captured nuclei |
standard_analysis |
Assigns nuclei to donors and detects doublets |
cell_classification |
Classifies cells based on their gene expression |
tensorqtl |
cis-QTL mapping using tensorQTL |