Wild-type viruses did not evolve to be medicines
AAV- based therapies are revolutionising the healthcare systems
and lives of millions of patients suffering from genetic
conditions.
Unfortunately, current generation of AAV
vectors based on naturally occurring capsids were not developed
for those therapeutic applications and transduce human tissues
with low efficiency.
This drives up the therapeutic
doses, which increases the cost and decreases the safety of
current AAV-based therapeutics.
Engineering novel AAV vectors with the desired biological
properties such as dossage efficiency, tissue-specificity or
immunogenicity is extremely challenging due to the huge space of
possible combinations to explore through iterations of directed
evolution.
It is a numbers problem!
A significant bottleneck of this technology which is
imposed by the bacterial transformation efficiency required to
expand the AAV plasmid library, which is in the order of 10 to 100
million clones.
As an example, if we want to explore a
7-mer peptide insertion, a complete randomisation
approach would encompas a theoretical sequence space of
1 Billion variants, and
10-mer insertion more than
10 Trillion variants!
Sendatu Therapeutics has developed the AAVitruvian platform, a propietary state-of-the-art capsid selection platform powered by artificial intelligence (AI) to speed-up the identification of optimal AAV vectors for the delivery of gene therapies to specific tissues.
Distinctive selection of ready-made capsids to identify best parent
capsid.
Functional transduction platform for parent capsid
modification – packaging, entry
and expression datasets.
We leverage cutting-edge
machine learning (ML) algorithms to train
models for capsid design. Our ML models can
screen billions of potential candidates and estimate their
efficiency with very high accruacy.
Models can be trained to predict how well the vector
will bind a specific cell in a tissue, or to
de-target an undesired tissue (e.g.
liver).
Balancing our model predictions and
previous data, we can design tailored AAV plasmid libraries
full of promising capsids.
Access to unique, biologically predictive,
human-relevant models.
Continuously
generating data through in-house lab expriments, which are used
to further train our machine learning models, creating a
feedback loop that keeps improving the
accuracy of the models.
Platform validated both in-vitro and in-vivo, showing an outstanding capacity to identify optimal capsid candidates with only two iterations of directed evolution (DE).
Sydney, New South Wales
Australia
contact@sendatutx.com