Computational Biologist · Research Software Engineer
An evolving collection of computational methods, models, and software for biological research.
Manuscript status: continuously under revision.
Email · LinkedIn · Google Scholar
Post-Doctoral Research Fellow · Centre for Regenerative Medicine “Stefano Ferrari” · University of Modena and Reggio Emilia
My work spans molecular simulation, protein engineering, structural bioinformatics and machine learning, with a particular interest in making computational models interpretable, reproducible, and useful to experimental scientists.
Structural Biology · Scientific Software · Molecular Modelling · Machine Learning · AI for Molecular Systems
Biological research increasingly relies on computational approaches to represent, model and analyse complex biological systems.
I work across different levels of biological representation, from molecular structures and sequences to computational and machine-learning models, developing both the methods and software needed to study them.
This perspective connects my work in structural bioinformatics, molecular simulation, protein engineering and AI.
Proteins · Nucleic acids · Molecular interactions · Genome editing
Structural bioinformatics & molecular modelling
Protein modelling · Molecular docking · Molecular dynamics · Structural analysis
Machine learning & deep learning
Deep learning · Geometric deep learning · Protein language models · Sequence-based models · Molecular representations · Explainability
Protein engineering
Computational design · Sequence analysis · Structure-based analysis
I develop and integrate computational methods across different stages of biological research, from molecular modelling and simulation to machine learning and scientific software. My work includes both methodological development and the implementation of practical workflows for biological applications.
Programming & development:
Python · C++ · JavaScript · Git · Linux
Molecular modelling:
PyMOL · Rosetta · OpenMM · GROMACS · MODELLER
Machine learning:
PyTorch · PyTorch Geometric · DGL · ESM
Biological & structural data:
PDB · UniProt · ChEMBL · BindingDB · KLIFS
Protein modelling and structural bioinformatics within PyMOL.
A platform integrating sequence analysis, structure prediction, homology modelling and structural analysis into a graphical molecular-modelling environment.
workflow design · tool integration · data handling · GUI development · automation
Molecular docking workflows within PyMOL.
A plugin-based environment for integrating and automating different docking protocols through a unified graphical interface.
plugin architecture · CLI integration · I/O standardisation · GUI workflows · automation
Protein structural analysis through contact-network representations.
A PyMOL plugin for constructing and analysing protein contact networks directly from molecular structures.
structural parsing · network construction · graph-based analysis · visualisation
A computational framework for allele-specific CRISPR/Cas design.
A web server and computational pipeline for designing allele-specific genome-editing strategies, from mutation analysis to guide ranking and selection.
sequence analysis · scoring · web server · gene editing
Deep learning for RNA G-quadruplex-binding protein prediction.
A sequence-based framework for predicting human proteins associated with RNA G-quadruplexes using deep-learning models.
protein classification · ESM embeddings · deep learning
Computational models are becoming increasingly powerful in biology, but predictive performance alone does not determine whether a model is useful to scientists.
My work increasingly focuses on the connection between biological representation, computational models and biological interpretation: understanding how molecular information is encoded, what models learn from these representations, and how computational methods can be made reproducible and practically useful.
A current research direction is the development of structure-informed deep-learning methods for molecular recognition and drug discovery, with a focus on kinase selectivity.
The Virtual Kinome Assay (VKA) is a framework designed to predict compound–kinase binding affinity across the human kinome by combining curated kinase structures, molecular docking and geometric deep learning. The aim is to provide a fast and scalable complement to experimental kinome profiling, while making predictions informative beyond a single affinity score.
A central component of the project is interpretability: gradient-based explainability methods will be used to identify the regions of kinase–ligand complexes that contribute most to predicted affinity, helping connect model predictions with molecular interactions, potency and selectivity.
kinome profiling · molecular docking · geometric deep learning · binding affinity · interpretable models
A second research direction focuses on AI-assisted protein design for selective targeting of difficult protein interfaces.
DYNAMITE aims to engineer a peptide that selectively recognises N-Myc while discriminating against the highly similar interfaces of c-Myc, L-Myc and MAX. The designed binder will form part of a peptide PROTAC intended to promote targeted N-Myc degradation in MYCN-amplified neuroblastoma.
The computational framework combines structural information, AI-driven inverse protein design and negative design, with biochemical and cellular experiments feeding back into successive design cycles. Explainable methods are integrated into the design process to identify the molecular determinants of selective recognition and generate experimentally testable mechanistic hypotheses.
protein design · inverse folding · negative design · deep learning · explainability · targeted degradation
You've made it this far. Thank you for reading the manuscript. The author welcomes correspondence, collaborations, and well-motivated pull requests.
Selected material from lectures, workshops and computational biology training.
Scripts and tools for bioinformatics, structural analysis, molecular modelling and scientific workflows.
Data and code availability: selected projects and scientific software are available through the repositories linked above.