Secondary Analysis¶

Secondary analysis takes raw sequencing reads (FASTQ files) and runs a GPU-accelerated variant calling pipeline powered by NVIDIA Parabricks. The pipeline calls SNVs and indels, with optional add-ons for copy number variation (CNV) and structural variant (SV) calling. All results are automatically loaded into AIVA for analysis, along with BAM files for visual review in IGV.

What the pipeline produces¶

Pipeline workflow¶

graph LR
    A[FASTQ Files] --> B[Parabricks Secondary Analysis]
    B --> C[BAM]
    C --> D[SNV/Indel Calling]
    C --> E[PGx Analysis]
    C --> J[CNV Calling]
    C --> K[SV Calling]
    D --> F[Annotation]
    F --> H[Ready for Analysis]
    E --> I[Star Alleles & Recommendations]
    I --> H
    J --> L[CNV Annotation]
    L --> H
    K --> M[SV Annotation]
    M --> H

1. Download: FASTQ files are downloaded from the provided cloud URLs.
2. Parabricks secondary analysis: Reads are aligned to the reference genome and small variants (SNVs and indels) are called using GPU-accelerated tools (DeepVariant for germline, DeepSomatic for tumor workflows). Read phasing is automatically enabled for germline and somatic paired pipelines, adding haplotype information to all variant calls. Aligned reads are saved as BAM files with IGV links for visual review.
3. Annotation + downstream analysis (parallel): Once Parabricks completes, multiple processes run in parallel:
   • Annotation: Small Variant Annotation enriches the output VCF, which is then parsed and loaded into the AIVA database.
   • PGx analysis: BAM files are used to assign pharmacogenomic star alleles, predict metabolizer phenotypes, and generate CPIC drug recommendations for 88 pharmacogenes. See Pharmacogenomics.
   • CNV calling (optional): Copy number variation analysis. Results are annotated and loaded into AIVA.
   • SV calling (optional): Structural variant calling. Results are annotated and loaded into AIVA.

You can monitor each stage in real time using the Job Manager.

Starting a pipeline run¶

Navigate to the Samples tab, click Upload, and select the Secondary Analysis tab.

Step 1: Name your sample¶

Enter a Sample Name. This is the name that will appear in your sample list and is used for @samples: mentions in AIVA Chat.

Step 2: Configure the pipeline¶

Optionally assign the sample to a Project for team collaboration.

Step 3: Provide FASTQ files¶

Enter the cloud storage URLs for your Read 1 (R1) and Read 2 (R2) FASTQ files. Supported URL schemes:

For paired-end sequencing, both R1 and R2 URLs are required. Click + Add Lane if your sample was sequenced across multiple flow cell lanes.

Step 4: Optional Interval File¶

• Interval File: Upload a BED file to restrict analysis to targeted regions (recommended for WES data).

Step 5: Start the pipeline¶

Click Start Pipeline to submit. The pipeline run is queued and processed in the background.

Somatic analysis¶

Somatic paired (tumor-normal)¶

Select Somatic Paired as the analysis type for matched tumor-normal samples. You will need to provide FASTQ files for both the tumor and normal samples, each labeled with the appropriate role.

Somatic tumor-only¶

Select Somatic Tumor-Only when a matched normal sample is not available. Only tumor FASTQ files are required.

Supported FASTQ formats¶

• Standard FASTQ format (.fastq, .fastq.gz, .fq.gz)
• Gzip-compressed files are recommended to reduce download time
• Both single-end and paired-end reads are supported

Troubleshooting¶

Next steps¶

• Monitor your pipeline job
• Explore the table view
• Explore your data in AIVA Chat
• Upload a VCF file directly