---
title: AAV design on the platform
description: >-
  Select an AAV capsid serotype for a target tissue and optimize the transgene
  cassette for AAV packaging — the gene-therapy tools of the biologics cluster.
tags:
  - aav
  - gene-therapy
  - biologics
---

An AAV gene therapy is two design problems in sequence: which capsid delivers your payload to the right tissue, and how to fit that payload inside the vector's tight packaging limit. The platform's AAV tools handle both — one selects a capsid serotype for a target tissue given tropism and pre-existing-immunity data, the other codon-optimizes a transgene cassette and checks it against the ~4.7 kb AAV packaging capacity. Together they are the gene-therapy sibling of the [antibody engineering](/docs/antibodies) cluster.

## Try it

**Try this prompt** — [open in a guest chat](/chat/new?guest=true&prompt=What%20does%20the%20literature%20report%20about%20AAV%20serotype%20tropism%20for%20liver%2C%20with%20the%20citation%20and%20the%20species%20each%20observation%20came%20from%3F%20Then%20codon-optimize%20the%20PCSK9%20protein%20sequence%20and%20tell%20me%20whether%20the%20construct%20fits%20the%20AAV%20packaging%20limit.).

> What does the literature report about AAV serotype tropism for liver, with the citation and the species each observation came from? Then codon-optimize the PCSK9 protein sequence and tell me whether the construct fits the AAV packaging limit.

_The serotype lookup then transgene optimization — the two AAV steps run in order from one prompt._

## The two AAV tools

**`aav_capsid_select`** looks up what the literature reports about AAV serotype tropism for a target tissue. For each serotype it returns the tissues a cited study reports, the qualitative level that study reports, the species or system it was observed in, and the citation — plus reported seroprevalence where a study exists. It does not score, rank or recommend a serotype, and the records come back in alphabetical order.

**`aav_transgene_optimize`** takes a transgene sequence and produces a codon-optimized cassette. It reverse-translates the sequence, optimizes codon usage against the human codon table with DNA Chisel, avoids a fixed set of restriction sites, and validates the payload size against the AAV packaging limit.

> **Neither tool designs a cassette, and neither takes a preference you might expect**

> `aav_capsid_select` takes a tissue, optionally one serotype to return in full, and optionally a payload size. There is no `species`, `route_of_administration` or `neutralizing_antibody_profile` input: species is reported per observation from the literature rather than requested, seroprevalence comes from the cited studies, and route is not modelled.
>
> `aav_transgene_optimize` optimizes the coding sequence only. There is no `promoter`, `polyadenylation` or regulatory-element input — those are cassette design choices the backend does not make, so assemble them yourself around the sequence it returns. Codon usage is fixed to *Homo sapiens*.

## Citations

> Zulkower V, Rosser S. DNA Chisel, a versatile sequence optimizer. *Bioinformatics*. 2020;36(16):4508-4509. doi: [10.1093/bioinformatics/btaa558](https://doi.org/10.1093/bioinformatics/btaa558)

## When to reach for them

Reach for **capsid selection** when you are:

- Starting an AAV gene-therapy design for a specific tissue (CNS, liver, muscle, retina).
- Comparing capsid serotypes for tissue tropism and immunogenicity.
- Selecting capsids with low seroprevalence in a target patient population.

Reach for **transgene optimization** when you are:

- Designing an AAV transgene cassette after capsid selection.
- Optimizing codon usage for human expression.
- Fitting a transgene within the AAV packaging capacity.

For a small-molecule or antibody program these are not the tools — see [antibody engineering](/docs/antibodies) or [cell & gene therapy](/docs/cell-therapy) for CAR-based modalities.

## How they fit together

The two tools run in order: select the capsid first, then optimize the transgene to fit it. Capsid selection tells you the packaging limit you are designing against; transgene optimization checks your cassette against that limit and reports whether it will package. Both are gene-therapy end steps — they produce a vector design rather than feeding another antibody tool.

## How they run

Both AAV tools run as asynchronous GPU jobs on Modal compute. You submit the design parameters, the platform tracks and polls the job, and a result card lands on the canvas when it finishes. Capsid selection typically takes a few minutes; transgene optimization can take longer, since codon optimization is an iterative search.

## Read next