Published tropism reports per serotype — tissue, level, model, citation — in alphabetical order with no score, plus the aav_packaging_feasible go/no-go.

Capsid selection returns published observations, not a verdict: per serotype, which tissues a cited study reports, the level that study reports, the species it was observed in, and the citation. Transgene optimization does return a verdict — aav_packaging_feasible, whether the optimized construct fits the packaging limit.

So the two results are read differently. The transgene result you can act on directly. The capsid result you have to read, because nothing in it ranks anything for you.

There is no composite score, and that is deliberate

An earlier version of this tool returned a composite_score that sorted serotypes by 0.70 * tissue_tropism_score + 0.30 * (1 - immune_prevalence). It was removed because the inputs were not real: the tropism numbers were hand-entered without sources, three of the nine seroprevalence figures were invented outright, and a caller's serotype_preference silently added to that serotype's score — a preference nudging its own answer. If you are carrying a composite_score, tissue_tropism_score or ranked_capsids from an older run or an older version of this page, it is not a measurement and nothing reproduces it.

Capsid selection result

Under the result key literature_serotype_lookup. Serotypes come back in alphabetical order; the ordering carries no meaning.

Field	Type	What it tells you
`method`	string	Always `literature_serotype_lookup`.
`query`	object	What was asked: the normalised `target_tissue`, `target_tissue_recognized`, and the `requested_serotype`.
`serotypes`	list	One record per serotype in the reference table (see below).
`n_serotypes`	int	How many records were returned.
`tissues_with_reports`	string[]	Every tissue any source in the table reports.
`packaging_limit_kb`	float	The ~4.7 kb limit applied.
`packaging_citation`	string	The source for that limit.
`payload_within_packaging_limit`	bool \| null	Whether your `max_payload_kb` fits. `null` when you sent no payload.
`disclaimer`	string	The result's own statement of what it is not.
`reports_for_tissue`	object \| null	The reports narrowed to the tissue you asked about.
`requested_serotype_record`	object \| null	The full record for your `serotype_preference`, when you set one.

Per-serotype record

Field	Type	What it tells you
`serotype`	string	The serotype.
`reported_tropism`	list	One entry per reported observation (see below).
`tissues_reported`	string[]	The tissues this serotype has any report for.
`seroprevalence`	object \| null	Reported anti-AAV total IgG prevalence, the population, and the citation. `null` when no study is cited — not zero.
`seroprevalence_note`	string \| null	A qualification on that figure.
`packaging_limit_kb`	float	Packaging capacity.
`packaging_citation`	string	Its source.
`caveats`	string[]	Recorded limitations for this serotype. Read them.
`payload_kb`	float \| null	The payload you asked to check.
`payload_within_packaging_limit`	bool \| null	Whether it fits this serotype.

Each reported observation

Every entry in reported_tropism:

Field	Type	What it tells you
`tissue`	string	The tissue the source reports.
`reported_level`	string	The level that source reports: `high`, `moderate` or `low`. A published qualitative call, not a computed number.
`model`	string	The species or system it was observed in, e.g. `human`, `mouse`, `c57bl6_mouse_only`.
`citation`	string	The source.

Reading the capsid result honestly

Three properties of this result change how you should use it:

Absence is missing evidence, not a low score. A tissue no source reports is simply absent from reported_tropism. It does not appear with a zero, so you cannot infer that the serotype fails there.
model is load-bearing. A high reported in c57bl6_mouse_only is not a claim about humans. Check the model on every entry you rely on before carrying it across species.
A missing seroprevalence is not a low one. null means no study is cited for that serotype.

Transgene optimization result

Field	Type	What it tells you
`optimized_sequence`	string	The codon-optimized DNA sequence.
`length_bp`	int	Length in base pairs.
`length_kb`	float	Length in kilobases.
`gc_content`	float	GC fraction.
`aav_packaging_feasible`	bool \| null	Whether `length_kb` fits `max_payload_kb` — the go/no-go.
`max_payload_kb`	float \| null	The packaging limit applied.
`protein_length_aa`	int	Input protein length in amino acids.
`all_constraints_pass`	bool	Whether every optimization constraint was satisfied.
`codon_optimization_score`	float	Objective score from the optimizer.
`species`	string	Codon table used — always `h_sapiens`.
`restriction_sites_avoided`	string[]	Restriction sites removed during optimization.

What to do next

Read caveats and the model on each reported_tropism entry before shortlisting a serotype. The result will not shortlist for you.
Check query.target_tissue_recognized. A false means no source in the table reports your tissue at all, so an empty reports_for_tissue says nothing about any serotype.
An aav_packaging_feasible: false means shortening the transgene or choosing a higher-capacity serotype. Codon optimization does not materially change length_kb, so trimming the input sequence is what brings it under the limit.
Promoter, polyA and regulatory-element choices are still yours — this tool optimizes the coding sequence only.

AAV outputs and interpretation

Capsid selection result

Per-serotype record

Each reported observation

Reading the capsid result honestly

Transgene optimization result

What to do next

Read next

AAV inputs

AAV design on the platform

On this page