Custom Bioinformatics Pipelines
End-to-end NGS, metagenomics/microbiome, and proteomics workflows: automated, scalable, and reproducible, with QC gates and versioned outputs.
Independent consulting practice of Marco Anteghini, PhD
Bioinformatics & ML, delivered.
Reproducible pipelines. Decision-ready outputs.
Mosaic Bioinformatics turns noisy biological and health data into reliable, reproducible results: custom pipelines, machine-learning models, and clear analysis for life-science research and industry.
About
A focused, senior bioinformatics partner
Mosaic Bioinformatics is the independent consulting practice of Marco Anteghini, PhD. It delivers end-to-end computational work, from raw sequencing reads and administrative records to validated models and reporting, built to be reproducible, documented, and ready to hand over.
Engagements typically span three domains:
- Biotech & life sciences: sequence/variant analysis, microbiome and proteomics pipelines.
- Healthcare & administrative data: SQL/ETL pipelines, cohort building, indicators, and QC.
- EU-funded R&I projects: technical contributions, structured documentation, and delivery support.
Services
What I deliver
Practical, validated computational work, scoped to your data and built to last beyond the engagement.
AI & Machine Learning
Models for protein property, variant impact, and risk prediction, using embeddings and protein language models, with explicit validation and failure modes.
Data Engineering & Analysis
SQL/ETL pipelines, cohort and indicator definitions, and reproducible extracts, plus dashboards and publication-ready figures.
Project & Delivery Support
Technical contributions and structured documentation for EU-funded consortia and R&D teams, with handover-ready artefacts and traceable decision logs.
Approach
From noisy inputs to decisions you can act on.
Every engagement is grounded in your data and your constraints, then built to be reproducible, validated, and ready to hand over.
- Reproducible by default: containerised, versioned, and documented.
- QC gates and explicit validation at every step.
- Interpretable outputs, with stated assumptions and limits.
- Clean handover: code, documentation, and decision logs.
1. Ingest & structure
Raw sequencing, variant, microbiome, or administrative data, standardised and quality-checked.
2. Pipelines & QC
Automated, reproducible workflows with QC gates and versioned, traceable outputs.
3. Model & validate
Machine-learning and embedding-based models, evaluated with explicit failure modes.
4. Decide & hand over
Interpretable results, reporting, and documentation ready for operational decisions.
Projects
Selected work
Representative engagements across research and industry. Client identities are kept confidential.
Pathogenicity Risk Assessment
Computational tools and ML workflows for predicting the impact of genetic variants on protein stability, structure, function, and localisation.
Data Management & ETL
Engineering and standardisation of complex administrative and biological datasets, with QC checks and reproducible extracts.
Gut Microbiota Analysis
Reproducible pipelines for taxonomic and functional profiling and dysbiosis detection in human gut microbiome data.
Sustainable Bio-Manufacturing
Computational biology and decision-support modelling for sustainable bio-manufacturing within EU-funded research consortia.
Credentials
Background & expertise
Academic foundation and applied experience behind the practice.
PhD in Bioinformatics
Deep-learning approaches for protein function, interaction, and localisation.
MSc Bioinformatics · BSc Biological Sciences
Strong grounding in molecular biology and computational methods.
Marie Skłodowska-Curie Fellow
H2020 MSCA-ITN fellowship (2018 cohort).
Co-founder, Bioinform (NGO)
Training schools and materials in bioinformatics and applied AI across Europe.
Host, BioIntelligent podcast
Conversations at the intersection of biology, data, and industry.
Reproducibility-first
Containerised, versioned, and documented work designed for clean handover.
Core toolkit
PythonSQLRBash
PyTorchTensorFlow
PostgreSQLDockerGit
Protein embeddings / pLMsVariant & mutation analysis
NGS & microbiome pipelines
Publications
Peer-reviewed research
A selection of published work in bioinformatics and machine learning.
Computers in Biology and Medicine · 2025
How did we get there? AI applications to biological networks and sequences
How AI drives biomedical science: processing large datasets and predicting complex phenomena.
View publication
Int. J. of Molecular Sciences · 2021
In-pero: deep learning for peroxisomal protein localization
Embedding-based deep learning to predict peroxisomal protein localization with high accuracy.
View publication
Computational & Structural Biotech. J. · 2022
OrganelX web server for sub-organelle protein localization
Precise sub-organelle protein localization using machine-learning approaches.
View publication
J. of Cellular Biochemistry · 2023
PortPred: transporter protein identification
Deep learning for identifying transporter proteins and predicting their substrates.
View publication
Doctoral Thesis · 2022
Deep learning for peroxisomal proteins
Revealing function, interactions, and localization using deep-learning approaches.
View publication
Blog
June 4, 2026
Latest articles
Notes on bioinformatics, machine learning, and reproducible data work, from me and occasional guest contributors.
Welcome to the Mosaic Bioinformatics blog
This is a space for practical writing on bioinformatics, machine learning, and reproducible data work.
Read article
Contact
Let's talk about your data
Whether it's a pipeline to build, a model to validate, or a project to support, get in touch to discuss scope, timeline, and how Mosaic can help.