01 / Question
Does copy number become functional overdosage?
Three genomic copies do not automatically imply a therapeutically meaningful increase in every neural cell state.
Founder · Longinus Therapeutics
From machine cognition to programmable medicine.
I am Marcio Diaz. My background is in software and machine cognition; I now lead Longinus Therapeutics, where we are exploring a narrow question in gene-dosage biology: whether partial reduction of CYFIP1 could be therapeutically useful in 15q11.2 BP1-BP2 microduplication.
Dosage system · conceptual
The engineering problem
Correct excess expression without crossing the lower boundary. In dosage biology, the target is a window—not zero.
Now building
Longinus Therapeutics · precision medicine for 15q11.2 duplication
Current work
A narrow biological question.
The duplication adds a genomic copy; the working hypothesis is that part of the phenotype may follow from excess CYFIP1 dosage. The difficult part is determining whether that chain holds in relevant human cells—and whether it can be adjusted safely.
02 / Constraint
The dose-response is probably not linear.
CYFIP1 loss is also harmful. Any useful intervention must move toward a physiological range without overshooting it.
03 / Tool
An antisense oligonucleotide.
The lead modality is an intrathecal ASO approach, with partial knockdown, transcript selectivity and CYFIP2 preservation treated as design constraints.
Selected writing
Selected writing.
Recent notes cover dose-response curves, ASO design, patient-derived neurons and genetic uncertainty. Earlier entries cover machine cognition and computational systems.
Therapeutics
ASO Design as Software Engineering
Why sequence-defined medicine feels structurally familiar to a software builder.
Dosage biologyThe U-Shape of CYFIP1 Dosage
The therapeutic window sits between two biological cliffs.
Career transitionFrom Machine Cognition to Programmable Medicine
Why sequence-defined medicine feels structurally familiar to a software builder.
Background
A change of substrate.
The questions moved from software to neural systems and then to molecular biology. The useful habits transferred; many of the intuitions did not.
Software systems
Earlier work in software development and applied research.
Machine cognition
Consciousness, interpretability and the architecture of internal representations.
Human genetics
Gene dosage, incomplete evidence and the limits of what a clinical report can establish.
Longinus
Building a precision-medicine programme for 15q11.2 duplication.
Contact
Questions, corrections and collaborations.
Email is open to scientists, clinicians, families and biotech operators working on dosage-sensitive neurodevelopmental disease. Technical corrections are especially welcome.
[email protected]