How to Rebuild the MCPH Protocol for 2026

Gene Identity and Alias Normalization

First we make a canonical gene table. Every later database depends on this. If the symbol layer is wrong, the whole analysis becomes quietly wrong.

Question answered

Are all 33 genes being recognized under their current and historical names?

Candidate sources

• HGNC REST API: approved symbols, previous symbols, aliases, HGNC IDs.
• Ensembl BioMart or REST: Ensembl stable IDs and genomic coordinates.
• NCBI Gene: Entrez IDs and older literature compatibility.

Expected output

current_symbol | bundle_symbol | aliases | HGNC_ID | Ensembl_ID | Entrez_ID
KNL1           | CASC5         | D40...  | ...     | ...        | ...
TRAPPC14       | C7orf43       | MAP11   | ...     | ...        | ...

This is the layer that prevents KNL1 and TRAPPC14 from disappearing just because an older database still knows them as CASC5 and C7orf43.

Clinical Gene-Disease Validity

This layer asks whether a gene is clinically supported as a microcephaly gene, not merely present in a historical list. It is the most important first update because it separates established MCPH genes from candidates, syndromic microcephaly genes, and genes that need cautious interpretation.

Question answered

Which of the 33 genes are currently strong, moderate, weak, or uncurated microcephaly genes?

Candidate sources

• PanelApp Severe microcephaly: green/amber/red clinical evidence and inheritance.
• Gene2Phenotype: allelic requirement, mutation consequence, mechanism, confidence.
• GenCC: harmonized gene-disease validity assertions from multiple curators.
• ClinGen: conservative expert validity curations where available.
• OMIM: phenotype names, inheritance, original gene-disease history; use with licensing awareness.

Plot proposal

A clinical validity heatmap with genes as rows and evidence sources as columns.

gene x evidence source
PanelApp: green / amber / red / absent
G2P: confirmed / probable / possible / absent
GenCC: definitive / strong / moderate / limited / disputed
ClinGen: definitive / strong / moderate / limited / no curation
OMIM: phenotype record present / absent

This plot would tell us which parts of Mario's list are clinically mature and which parts are still research-facing.

Phenotype Ontology and Disease Vocabulary

Disease labels are messy. “Primary microcephaly,” “severe microcephaly,” “microcephalic dwarfism,” “cortical malformation,” “neurodevelopmental disorder with microcephaly,” and “intellectual disability with microcephaly” overlap but are not identical. Before plotting cross-condition intersections, we need a controlled vocabulary.

Question answered

What phenotype or disease entity is each database actually talking about?

Candidate sources

• HPO: phenotype terms such as microcephaly, seizures, intellectual disability, cortical malformation.
• MONDO: disease entities and cross-database mappings.
• Orphadata: rare disease classifications and Orphanet identifiers.

Plot proposal

A phenotype map or nested disease-tree annotation beside the gene table.

This layer protects us from treating every database hit containing the word “microcephaly” as if it meant the same biological condition.

Variant, Inheritance, and Constraint Evidence

A gene can be associated with disease in different ways: biallelic loss-of-function, dominant-negative variants, dosage sensitivity, missense-specific mechanisms, or uncertain candidate variants. This layer asks what kind of genetic damage the gene tolerates and what variant classes are reported in patients.

Question answered

What variant logic explains each gene-disease relationship?

Candidate sources

• ClinVar: pathogenic and likely pathogenic variants, with review status caveats.
• gnomAD: LOEUF, pLI, missense Z, observed/expected constraint.
• ClinGen dosage sensitivity: haploinsufficiency and triplosensitivity where curated.
• DECIPHER: patient-level phenotype-genotype evidence where accessible.

Plot proposal

gene | inheritance | variant class | LOEUF | pLI | missense Z | ClinVar P/LP count | dosage sensitivity

This layer helps us distinguish “this gene is generally important” from “this kind of mutation in this gene causes this kind of microcephaly.”

Developmental Brain Expression

This is the neuroscience heart of the update. The old bundle includes developmental expression blocks, but modern brain atlases can ask more precise questions: which cortical progenitors express these genes, at what developmental stages, and how do expression profiles differ between radial glia, intermediate progenitors, neurons, organoids, and non-human primate comparisons?

Question answered

When and where in brain development are these genes likely to matter?

Candidate sources

• BrainSpan: foundational human developmental brain transcriptome downloads.
• NeMO Analytics: recent neocortical development compendium, highly relevant to Gabriel Santpere's line.
• PsychENCODE: neuropsychiatric brain regulatory and transcriptomic resources.
• CZ CELLxGENE Census: programmatic single-cell access in Python and R.

Plot proposal

genes x developmental cell states
apical radial glia
basal radial glia
intermediate progenitors
early neurons
cortical plate neurons
organoid progenitors
non-human primate / human comparative stages

This layer would transform the atlas from a gene-list exercise into a developmental neuroscience argument.

Cross-Condition Genetic Associations

The old plot already points in this direction with MAGMA and GWAS-like disease columns. The modern version should be more explicit about what kind of evidence is being shown: Mendelian gene causality, common-variant association, locus-to-gene prioritization, literature co-mention, pathway evidence, or functional regulatory evidence.

Question answered

Do MCPH genes intersect with ASD, schizophrenia, cognition, ADHD, bipolar disorder, Alzheimer disease, Parkinson disease, or other brain-related traits?

Candidate sources

• Open Targets Platform: integrated target-disease associations through GraphQL and downloads.
• NHGRI-EBI GWAS Catalog: curated GWAS associations and REST API.
• PGC cross-disorder resources: psychiatric common-variant signals and gene-set analyses.
• FinnGen, UK Biobank, GeneBass: broader disease/trait associations where mapping is appropriate.

Plot proposal

gene x condition
Open Targets association score
GWAS Catalog mapped traits
MAGMA or gene-based signal
PGC membership
old bundle MAGMA comparison

This layer must be labeled carefully: GWAS evidence does not mean “this MCPH gene causes schizophrenia.” It means the gene may sit near, or be prioritized from, broader trait-associated biology.

Mechanism, Pathway, and Network Interpretation

Once clinical and association evidence is mapped, we need to ask what the genes do together. This is where the project becomes understandable for a neuroscience reader: not 33 names, but a set of biological modules that explain why neural progenitors are vulnerable.

Question answered

What biological systems organize the 33 genes?

Candidate sources

• Gene Ontology: biological process, cellular component, molecular function.
• Reactome: curated pathways and reactions.
• STRING and BioGRID: protein interaction networks.
• g:Profiler and Enrichr: enrichment analyses.
• MSigDB and KEGG: useful but check licensing/redistribution constraints.

Plot proposal

A module map: centrosome, centriole, kinetochore, spindle checkpoint, cytokinesis, nuclear envelope, DNA repair, chromatin regulation, RNA/ribosome biology, trafficking, autophagy, lipid transport, and BBB biology.

This is the interpretive layer that lets the thesis say: “microcephaly genes converge on progenitor proliferation, genome integrity, and developmental timing.”