Skip to content

Platform overview

AIVA covers the full journey from raw sequencing data to a signed clinical report. This page walks through each stage so you know what the platform does and where to find it.

Pipeline
FASTQ Variant Calling BAM PGx Annotation VCF GPU PIPELINE DIRECT UPLOAD
Step 1

Upload your data

Everything starts with a file. Upload FASTQ files and let GPU-accelerated Parabricks pipelines call variants, generate BAMs, and assign PGx star alleles. Or skip the pipeline: drop a VCF and AIVA annotates it directly.

  • Small variant calling powered by NVIDIA Parabricks on GPUs
  • BAM file generation with direct IGV links for visual review
  • Pharmacogenomic variant calling and star-allele assignment
  • Automated annotation with ClinVar, gnomAD, and DITTO scoring
Samples & Uploads →
AIVA Chat
List P/LP variants in @sample:patient007
AIVA Found 12 pathogenic variants. Showing top results by classification…
GeneVariantClass
BRCA1c.5266dupC (p.Gln1756fs)Pathogenic
TP53c.718C>T (p.Gln240*)Pathogenic
MLH1c.1972dupC (p.Gln658fs)Likely Path.
Step 2

Ask AIVA Chat

Once your data is loaded, AIVA Chat acts as your AI Clinical Analyst. Ask a question in plain English and get classified, evidence-backed answers grounded in your sample data.

  • Review and classify variants with a single prompt
  • Automated literature review with cited evidence
  • HPO-driven gene prioritization
  • Knowledge graph associations between genes, diseases, and drugs
AIVA Chat →
Variant Table
GeneHGVSDITTOClass
BRCA2c.5946delT0.97 Path
PTENc.388C>T0.91 LP
APCc.3920T>A0.54 VUS
RB1c.2117G>A0.42 VUS
whole genome scale · real-time filtering
Step 3

Analyze variants manually

Whole genome sequencing produces millions of variants per sample. The data table lets you filter by gene, consequence, frequency, or ACMG class and see results in real time.

  • Virtualized rendering with zero performance degradation at whole-genome scale
  • One-click export of filtered views to CSV, TSV, or direct to report
  • Per-variant flagging, comments, and threaded review for clinical sign-off
Data Table →
Clinical Report
Interpretation
Findings
Pathogenic
VUS
Step 4

Generate a clinical report

Pull your findings into a clinical report. AI-assisted auto-fill populates report sections from your variant data and analysis findings.

  • AI Auto-Fill drafts interpretations automatically
  • Edit and refine auto-generated content
  • Upload custom report templates per assay type
  • Sign and export reports
Reports →

Putting it all together

flowchart LR
    A[FASTQ / VCF] --> B[Annotation]
    B --> C[AIVA Chat]
    B --> D[Data Table]
    C --> E[Clinical Report]
    D --> E

Every module connects. Upload a file, analyze it with chat or the data table (or both), and produce a report. No manual data handoff between tools.

For collaboration features like projects, role-based access, and variant flagging, see Collaboration. For automating workflows, see Playbooks.