Draft route · not final
Six weeks, from school science to the state of the field.
A route for a reader with school science: about six weeks of evenings. Each step names a tool, how long it takes, and what you can explain afterwards. Where no tool exists, the step says so.
2 of 26 steps open a tool and 9 more have records to read. The remaining 15 are gaps, shown dashed: they are the build list.
This is a draft
This order is a first proposal by the session that keeps the atlas, for Amit to change. Times are estimates; nobody has yet walked a real high schooler through it. A step with records to read has no tool yet: the records are the reading.
Week 1
What a cell is made of
Start with the smallest thing that is alive, and count its parts.
- 40 minEvery Gene: play Shed or keep
Afterwards you can explain: what it means for a to be essential, and how a knock-out tests it.
- 1 hourEvery Gene: read ten gene cards at the first two depths
Afterwards you can explain: the difference between a measured, an inferred and a simulated statement.
- readingDNA, RNA and protein from the ground up
Read: The 5' and 3' ends, RNA capping and the poly(A) tail, Alternative start codons and two protein forms.
Afterwards you can explain: how a sequence becomes a , and what can go wrong at each step.
No tool yet. Closest thing that exists: nothing in the registry starts this low; the paused Mechanism Ladder series is the nearest.
- readingThe chemistry underneath: bonds, water, energy
Read: Thermal energy (kT) as a yardstick, The hydrogen bond (in proteins and water), The hydrophobic effect, Salt bridge, Pi-pi stacking and cation-pi interactions.
Afterwards you can explain: why molecules stick, fold and react at body temperature.
No tool yet. Closest thing that exists: the OMSF concept reports (held, written for specialists).
Week 2
Reading the genome and its switches
How do we read the letters, and how do we see which parts are in use?
- readingHow sequencing works: short reads, long reads, variant calls
Read: Whole-genome sequencing, Whole-exome sequencing, Germline and somatic variants.
Afterwards you can explain: how a machine turns into letters, and how sure each letter is.
No tool yet. Closest thing that exists: Where the Data Is (held) says what each assay reads and cannot see.
- nothing yetGap: The epigenome: methylation, open chromatin, 3D folding
You would be able to explain: why two cells with the same behave differently, and how ATAC-seq and ChIP-seq show it.
Closest thing that exists: Where the Data Is (held).
- 1 hourVariant mechanisms: 96 words: the dosage and expression families
Afterwards you would be able to explain: a dozen ways a change in alters how much of a product is made.
This tool is built but not published yet, so this step is not available here.
- nothing yetGap: RNA: bulk, single-cell and spatial RNA-seq
You would be able to explain: how counting RNA molecules tells you what a is doing, and what the counts hide.
Closest thing that exists: VCC learning guide (held: challenge dashboard).
Week 3
Proteins: amount, shape, partners
Most of the work in a cell is done by proteins. How do we count them, see their shape, and find what they bind?
- readingMass spectrometry: weighing molecules
Read: Affinity purification mass spectrometry (AP-MS), Crosslinking mass spectrometry (XL-MS).
Afterwards you can explain: how a mass spectrometer identifies and counts .
No tool yet. Closest thing that exists: an MLDD mechanism report on fragmentation and the Phare mass-spec bench lab (both held).
- 1.5 hoursVariant mechanisms: 96 words: the fold, binding and modification families
Afterwards you would be able to explain: how one amino-acid change can break a fold, lose a partner or remove a chemical tag.
This tool is built but not published yet, so this step is not available here.
- readingStructure: crystallography, cryo-EM, NMR, predicted structures
Read: Structural evidence of complexes (crystallography and cryo-EM), Alpha helices and beta sheets.
Afterwards you can explain: how each method arrives at a shape, and what a confidence colour on a predicted structure means.
No tool yet. Closest thing that exists: Zika Protease Atlas (held: challenge dashboard) and the OMSF crystallography report (held).
- readingInteractions: who binds whom
Read: Yeast two-hybrid (Y2H), Affinity purification mass spectrometry (AP-MS), Proximity labelling (BioID, TurboID, APEX), Surface plasmon resonance (SPR), Binding kinetics: on rate, off rate and residence time, Has every protein been tested against every other protein?.
Afterwards you can explain: how pull-downs and proximity labelling map which work together.
No tool yet. Closest thing that exists: none.
Week 4
Whole cells and early bodies
Put the parts together. Can we predict what a whole cell, or an embryo, will do?
- 45 minMinimal Cell Worlds
Afterwards you would be able to explain: what a whole-cell contains, and how it is tested against measured growth and protein counts.
This tool is built but not published yet, so this step is not available here.
- nothing yetGap: Seeing inside cells: light and electron microscopy
You would be able to explain: how a microscope image is formed, and how we know where a lives.
Closest thing that exists: Cell Map Field Guide (held: challenge dashboard, already public).
- 45 minEmbryo Simulator: zygote to blastocyst
Afterwards you would be able to explain: how position inside an embryo decides a cell's first fate, and how a model's knock-outs are checked.
This tool is built but not published yet, so this step is not available here.
- nothing yetGap: Cell types and states: single-cell atlases, lineage tracing, flow cytometry
You would be able to explain: how we decide two cells are the same type.
Closest thing that exists: none.
- nothing yetGap: Signalling between cells
You would be able to explain: how a signal outside a changes what happens inside.
Closest thing that exists: none in the public cut.
Week 5
Intervening: drugs, knock-outs, trials
The strongest test of understanding is to change something and predict what happens.
- readingPerturbation: CRISPR screens and compound assays
Read: Hit discovery, High-throughput screening, Genetic interaction and epistasis screens, Luminescent cell viability assay (CellTiter-Glo), Dose-response assay, High-content imaging and morphological profiling.
Afterwards you can explain: how thousands of are tested at once, and what a dose-response curve shows.
No tool yet. Closest thing that exists: the paused Assay Lab series (held).
- nothing yetGap: What the body does to a drug (ADMET)
You would be able to explain: why a molecule that works in a dish can fail in a person.
Closest thing that exists: Phare Intelligence Atlas (held); the paused Pharmacology from the Body Up series.
- 1.5 hoursTrial postmortem: front page and the trials section
Afterwards you would be able to explain: what a measures, how many have been run, and how their quality changed over time.
This tool is built but not published yet, so this step is not available here.
- 1.5 hoursTrial postmortem: six cases from the ledger
Afterwards you would be able to explain: why a target that looked right in animals failed in people, in specific cases.
This tool is built but not published yet, so this step is not available here.
Week 6
Populations, evolution, and writing it all down
Step back: variation across people and species, the statistics that keep us honest, and how knowledge is stored for machines.
- readingHuman genetics: GWAS and biobanks
Read: GWAS and variant prioritisation, Genome-wide association study (GWAS).
Afterwards you can explain: how comparing many people's points to the causes of disease.
No tool yet. Closest thing that exists: the postmortem's evidence section touches this; no dedicated tool.
- nothing yetGap: Evolution as evidence: comparing species
You would be able to explain: what a lost in one species tells you about its job.
Closest thing that exists: Darth Genome (held: may hold unpublished work); the paused Evolution Mechanics series.
- nothing yetGap: Statistics you need to read a paper
You would be able to explain: what an interval, a p-value and a base rate do and do not tell you.
Closest thing that exists: Defensible newsletter (held: email only).
- nothing yetGap: Knowledge graphs and ontologies
You would be able to explain: how facts from thousands of papers are stored so a program can reason over them.
Closest thing that exists: Knowledge Graph Field Guide (held, already public).
- readingState of the field: what is solved, what is open
Read: Clinical, Models of models.
Afterwards you can explain: which big questions in biology are answered, which are being worked on, and where the data is thin.
No tool yet. Closest thing that exists: Where the Data Is (held).