← All research

Simulation study · Downstream processing

When is split AAV loading worth the extra cycle?

In an AAV2 purification model, split loading raised simulated recovered vector per liter while adding processing time. The useful decision is conditional: measure overload losses and real skid timing before choosing a strategy.

The decision

An overloaded chromatography step may leave valuable vector in flow-through, wash, shoulder, or strip fractions. Splitting the load might recover more vector from the same feed, but it can also add a cycle and change purity or cost. A purification team needs to know whether that extra cycle pays back under its own resin and skid conditions.

Ayomic compared a single load, split load, and adjusted product cut for an AAV2 reference case. The model held feed, flow, and total loading time equal across strategies, then added the strategy-specific non-loading phases.

A conditional trade-off

Across the modeled overload hypotheses, split loading produced 30–50% more recovered vector per liter of feed than the single-load comparison, with about 4.3% more total process time. It was the lowest-cost option in about half the sampled input draws, but the cost bands overlapped. The study did not find a universal winner.

Corrected modeled total time, bars scaled from 0 to 15 hours. All options share 13.33 hours of loading at the same feed and flow; the remaining time differs by strategy.
Corrected model timing at equal feed and flow
Strategy Loading Total modeled time
Single load 13.33 h 13.93 h
Split load 13.33 h 14.53 h
Adjusted cut 13.33 h 14.23 h

The corrected timing matters: a previous split-load calculation counted loading time twice and overstated its time penalty. The figures here use the revised, equal-feed comparison.

What the model tested

The study started from a published AAV2 purification reference, built three mass-balanced hypotheses for vector fate under overload, and compared a chromatography surrogate with a Python cost calculation. It explored how the strategy ranking changed when the assumed overload loss, product cut, and cycle overhead changed.

The next useful experiment is direct: at high load, measure vector and capsid distribution across feed, flow-through, wash, product, shoulder, and strip fractions. Record actual skid phase durations at the same time. Those measurements would replace the assumptions that currently drive the strategy choice.

What still needs proof

  • The overload behavior was simulated, not measured in a physical chromatography run.
  • The comparison applies only within the modeled range of roughly 0.5–2× reference loading and 0.6–0.92 product purity.
  • Resin, serotype, infectivity, and site-cost transfer need separate validation. No infectivity improvement is claimed.

These are conditional AAV2 model results, not a validated purification recipe or a demonstrated manufacturing gain.

Sources

Study snapshot: 3 October 2026. The study used public literature as an AAV2 baseline and chromatography plausibility context: Biotechnology and Bioengineering reference and Journal of Chromatography A reference. Neither paper validates the simulated strategy ranking shown here.

See how Ayomic approaches simulation and comparison