Wiki

The teaching layer under the posts — organized by topic, with a terminology index. Each idea is explained once here, so posts can link to it instead of re-explaining. The posts are interest-driven chapters; this is the reusable ground beneath them.

Terminology index

A-Lab → Self-driving lab
adaptation → Fine-tuning
AF2 → AlphaFold
AF3 → AlphaFold
AlphaFold
AlphaFold2 → AlphaFold
AlphaFold3 → AlphaFold
AlphaMissense → Variant-effect prediction
anonymization → De-identification
Attention
attention mechanism → Attention
autonomous experimentation → Self-driving lab
autonomous lab → Self-driving lab
base editing → Gene editing
base model → Foundation model
binder design → Protein design
biomedical knowledge graph → Knowledge graph
Cas9 → Gene editing
central dogma → The central dogma
clinical foundation model → EHR foundation model
CRISPR → Gene editing
de novo protein design → Protein design
De-identification
denoising diffusion → Diffusion model
Differential privacy
Diffusion model
diffusion models → Diffusion model
DNA to RNA to protein → The central dogma
docking → Molecular docking
DP → Differential privacy
EHR foundation model
EHR transformer → EHR foundation model
embedding → Embeddings
embedding space → Latent space
Embeddings
ESM → Protein language model
feature vector → Embeddings
Federated learning
Fine-tuning
finetuning → Fine-tuning
FL → Federated learning
FM → Foundation model
force spectroscopy → Single-molecule biophysics
Foundation model
Gene editing
gene expression → The central dogma
Geneformer → Single-cell foundation model
generative protein design → Protein design
in-silico perturbation → Perturbation prediction
KG → Knowledge graph
Knowledge graph
knowledge graphs → Knowledge graph
Large language model
large language models → Large language model
latent → Latent space
latent representation → Latent space
Latent space
LLM → Large language model
Molecular docking
mRNA vaccine
mRNA vaccines → mRNA vaccine
optical trap → Single-molecule biophysics
pathogenicity prediction → Variant-effect prediction
Perturb-seq → Perturbation prediction
perturbation modeling → Perturbation prediction
Perturbation prediction
PHI scrubbing → De-identification
PLM → Protein language model
pretrained model → Foundation model
prime editing → Gene editing
privacy budget → Differential privacy
privacy filtering → De-identification
Protein design
Protein language model
protein language models → Protein language model
protein-ligand docking → Molecular docking
RNA vaccine → mRNA vaccine
robot scientist → Self-driving lab
scGPT → Single-cell foundation model
self-attention → Attention
Self-driving lab
self-driving labs → Self-driving lab
single-cell FM → Single-cell foundation model
Single-cell foundation model
single-molecule → Single-molecule biophysics
Single-molecule biophysics
The central dogma
transfer learning → Fine-tuning
Transformer
transformer architecture → Transformer
transformers → Transformer
variant effect → Variant-effect prediction
Variant-effect prediction
vector representation → Embeddings
VEP → Variant-effect prediction
virtual cell → Single-cell foundation model

AI & Machine Learning

How models learn, represent, and reason.

AttentionconceptEach token deciding which other tokens to look at, weighted by relevance.
Diffusion modelconceptA generative model that learns to turn pure noise into a sample by reversing a gradual noising process, one denoising step at a time.
Embeddingsdeep-diveVectors of numbers that place data in a learned space, so math on the vectors reflects real relationships.
Fine-tuningconceptTaking a model that already learned a lot and nudging it to master one specific task.
Foundation modelconceptA large model pre-trained on broad data once, then reused for many downstream tasks.
Large language modelconceptA transformer trained on huge amounts of text to predict the next token, which yields broad language and reasoning ability.
Latent spaceconceptThe compressed internal 'map' a model learns, where similar things sit close together.
TransformerconceptA neural network that reads a whole sequence at once and lets every element decide which others matter.

Biology

Molecules, cells, and how life is organized.

The central dogmaconceptHow genetic information flows: DNA is transcribed to RNA, which is translated to protein.

Measurement & Inference

Seeing and quantifying — biophysics, statistics, causality.

Single-molecule biophysicsconceptWatching and measuring one molecule at a time instead of averaging over billions.

Bio × AI Methods

Where the two meet — structure, sequence, cells, molecules.

AlphaFoldconceptA model that predicts a protein's 3D structure directly from its amino-acid sequence.
Knowledge graphconceptA network of biomedical facts — genes, drugs, diseases, proteins linked by typed relationships you can query and reason over.
Molecular dockingconceptPredicting how a small molecule fits into a protein's pocket — the pose — and scoring how well it binds.
Perturbation predictionconceptPredicting how a cell's gene expression changes in response to a perturbation — without running the experiment.
Protein designconceptBuilding new proteins that don't exist in nature to do a specific job — the inverse of structure prediction.
Protein language modelconceptA language model trained on protein sequences, treating amino acids like tokens to learn the 'grammar' of proteins.
Single-cell foundation modelconceptA transformer pretrained on millions of single cells that turns a cell's gene expression into a reusable representation.
Variant-effect predictionconceptPredicting whether a genetic mutation is harmful or harmless — often from evolution and protein language models alone.

Domains & Disease

Proteins, genomes, cancer, the heart, aging.

Gene editingconceptMaking precise, targeted changes to a genome's DNA sequence.
mRNA vaccineconceptA vaccine that delivers mRNA instructions so your own cells make the target protein — you ship the recipe, not the protein.

Translation

Lab → clinic → society; trust and validation.

De-identificationconceptStripping the identifiers that tie a health record to a person — necessary to share clinical data, but not enough to make it anonymous.
Differential privacyconceptA mathematical guarantee that a result barely changes whether or not any one person's data was used.
EHR foundation modelconceptA model pretrained on millions of longitudinal patient records, then fine-tuned to predict what happens next in the clinic.
Federated learningconceptTrain one shared model across many sites without moving the raw data — the model travels to the data.

Frontier

Agentic science, self-driving labs, the virtual cell.

Self-driving labconceptA lab where AI proposes experiments, robots run them, and results feed back — closing the loop with minimal human intervention.