From what a PPI actually is, to running your own network analysis in STRING and Metascape — this guidebook takes you through the full picture, step by step, with real neuroscience examples.
A protein–protein interaction (PPI) occurs when two or more protein molecules bind to each other and, as a result, affect the function of one or both.
Proteins don't work alone. Nearly every biological process — from how a neuron fires an action potential to how a cell decides to die — depends on proteins forming physical contacts with other proteins. A single hub protein in the brain might interact with dozens of partners, and disrupting just one of those interactions can cascade into disease.
The complete set of PPIs in an organism is called the interactome. Mapping it tells us how biological information flows, which proteins are functionally important, and where disease processes can be targeted.
Direct, physical binding — like a key in a lock. Shape-specific.
Hub proteins connect to many partners, like a major airport connecting cities. Remove the hub — the network collapses.
Many PPIs pass a molecular message down a chain — each protein handing off to the next.
Two proteins physically touch — detectable by methods like yeast two-hybrid (Y2H) or co-immunoprecipitation. In STRING, these are sourced from databases like BioGRID and IntAct. Example: SNCA (α-synuclein) directly binds to PARK7 (DJ-1) in a protective complex relevant to Parkinson's disease.
Two proteins are found in the same complex but may not directly touch — detected by co-purification methods like AP-MS (affinity purification–mass spectrometry). Important in STRING's "co-expression" and "databases" channels. Example: APP and PSEN1 are found together in the γ-secretase complex along with nicastrin and APH-1, even though they don't all directly contact each other.
Stable interactions (e.g. ribosome subunits) are permanent structural associations. Transient interactions are fleeting — a kinase docking briefly to phosphorylate its target before leaving. Most signalling PPIs are transient, making them harder to detect experimentally but functionally critical. STRING captures both types across different evidence channels.
When you query STRING or Metascape, the tool doesn't distinguish what type of interaction it shows unless you filter by evidence channel. Understanding interaction types helps you interpret confidence scores and pick the right experimental follow-up.
Many of the most devastating neurological diseases are fundamentally diseases of disrupted protein interactions. Here are four major examples you'll encounter throughout this guide.
Amyloid Precursor Protein (APP) interacts with BACE1, PSEN1, and APOE in a network that, when disrupted, leads to Aβ plaque accumulation. PPI network analysis was instrumental in identifying these connections and proposing therapeutic targets.
SNCA forms a dense interaction hub with PARK2, PINK1, and LRRK2. These proteins all converge on mitochondrial quality control and ubiquitin–proteasome pathways. Network analysis revealed this convergence, shaping our understanding of PD pathophysiology.
The postsynaptic density (PSD) is one of the most interaction-dense structures in biology — hundreds of proteins interlocking to coordinate glutamate receptor trafficking, long-term potentiation, and plasticity. Mapping its PPI network revealed how psychiatric disease mutations converge.
BDNF binds TrkB (NTRK2), triggering a cascade of phosphorylation PPIs through PI3K/AKT and MAPK/ERK pathways. GO enrichment of genes in this network consistently highlights terms like "axon guidance" and "synaptic plasticity" — exactly the kind of result you'll learn to interpret.
Each chapter builds on the last. Start at chapter 1 and work through, or jump to any topic you need.
What proteins are, how they interact, what a network actually means, and why hub proteins matter so much in neuroscience.
STRING, BioGRID, IntAct, and MINT — what they contain, how evidence is scored, and how to read what they return.
Gene Ontology, the three namespaces, enrichment testing, and how to use g:Profiler and Metascape to interpret a gene list.
From Metascape's built-in networks to Cytoscape for publication-quality figures — metrics, community detection, and visualisation.
Three real neuroscience papers dissected step by step — what the authors did, why, and what you'd do to replicate their analysis.
Threshold selection, multiple testing correction, publication bias, and how to report PPI results so reviewers take them seriously.
You'll encounter these throughout the guide. Search below to find any term.
Chapter 1 covers everything you need to understand before touching any tool.