About Marcio

A software background, a genetic question and a lot still to learn.

I came to biology from software and machine cognition. A diagnosis in my family made one abstract question concrete: can a dosage-sensitive molecular system be adjusted precisely enough to matter?

The through-line

I began in software and applied research, later focusing on machine cognition, interpretability and theories of consciousness. I am not a clinician or an academic biologist; my contribution is to build the programme, ask useful questions and bring the right specialists together.

The direction changed when a clinical report in my family identified a recurrent copy-number variant. That led me into gene dosage, RNA biology and human cell models. The report named an interval; it did not identify which gene mattered, in which cell type, or whether changing it would help.

In 2026 I founded Longinus Therapeutics in Madrid. The company is focused on 15q11.2 BP1-BP2 microduplication and a bounded hypothesis: CYFIP1 overdosage may contribute to the biology, and controlled reduction may be useful. Both parts remain to be established.

Current work at Longinus

01

Human biology. Patient-derived iPSC models and neuronal systems that bring the programme closer to the biology of the people it is intended to serve.

02

A dosage-correction window. The objective is controlled normalization—not knockout—with CYFIP2 selectivity and safety treated as first-order design constraints.

03

ASO discovery. Sequence design and screening for potency, selectivity and safety, connected to the biology the programme is intended to change.

04

A Madrid operating base. A small internal discovery team using specialist cores and external regulated infrastructure instead of building every capability from scratch.

Approach

Longinus is being built as a small, focused programme. Patient-derived cell models, quantitative protein work and ASO screening require different kinds of expertise, so the plan combines a small internal team with specialist laboratories and regulated external infrastructure.

Software analogies are useful for sequence design and iteration, but biology is not deterministic, cells are not unit tests and a clean abstraction can still fail in an organism. The earlier machine-cognition archive remains here as a record of the work that preceded that lesson.

Contact

You can reach me at [email protected]. Contact is welcome from scientists, clinicians, families, platform builders and biotech operators working on gene dosage or neurodevelopmental disease.