Your First Workflow
In this tutorial you’ll build a small but realistic germline variant-calling pipeline: take paired-end reads, align them to a reference, and call variants. The same graph runs on the desktop or in the cloud.
The pipeline
[FASTQ Input (R1/R2)]
│
▼
[FastQC] ───────────────► [MultiQC report]
│
▼
[BWA-MEM] ◄── [Reference FASTA]
│
▼
[SAM → sorted BAM]
│
▼
[Mark Duplicates]
│
▼
[GATK HaplotypeCaller] ◄── [Reference FASTA]
│
▼
[VCF Output]Step 1 — Inputs
- Add a FASTQ Input node and set the R1/R2 read files.
- Add a Reference FASTA node and select (or upload) your reference genome. For a quick test, use the bundled E. coli or chr20 subset.
Step 2 — Quality control
Add a FastQC node and connect the FASTQ input to it. Optionally add a MultiQC node downstream to aggregate the report. QC never blocks the rest of the pipeline — these nodes run in parallel with alignment.
Step 3 — Alignment
- Add a BWA-MEM node.
- Connect the FASTQ reads to its
readsinput and the reference to itsreferenceinput. - In the inspector, set the read-group string (
@RG) — required by GATK downstream.
Step 4 — Post-processing
- Add a Sort & Index BAM node after BWA-MEM.
- Add a Mark Duplicates node (Picard/GATK) after the sort.
Step 5 — Variant calling
- Add a GATK HaplotypeCaller node.
- Connect the de-duplicated BAM and the reference.
- Connect its output to a VCF Output node.
Step 6 — Choose resources and run
Variant calling on a real genome is memory-intensive. In the run bar:
- Pick Medium (64 GB) or larger for whole-genome data; Small (16 GB) is fine for the chr20 subset.
- Click Run and watch the nodes execute in topological order.
Tip: If a long run is interrupted, the cloud engine resumes from the last completed node thanks to checkpointing — you don’t re-pay for work already done.
Step 7 — Save & share
Your workflow is stored as JSON. Use Export to download the workflow JSON, or share it with teammates on a Team-tier plan.