Foundations

AlphaFold 3 predicts biomolecular complex structures but reports limitations including clashes, stereochemistry errors and hallucinated regions.

RFdiffusion generates proposed protein backbones and constrained designs.

In the EX2 opening model, exchange rates constrain approximate opening free energies.

Ferrari and colleagues profile 5,778 small domains with multiplexed HDX-MS and compare 13 domains with site-resolved HDX NMR.

Turing proposed that reacting and diffusing morphogens can destabilize an initially homogeneous equilibrium and generate patterned states.

Traction microscopy infers forces from cell-induced substrate displacements using a substrate mechanics model.

Predicting conformational free energies, kinetics and function from sequence/static structures remains an open research challenge.

IGoR infers generative V(D)J models by considering alternative gene choices, deletions and insertions for each receptor sequence.

A receptor’s abundance or generation probability is not itself its antigen affinity.

Calcium fluorescence is an indirect activity proxy, and spike inference depends on indicator dynamics, noise and the trained/calibrated model.

Rupprecht and colleagues benchmark GCaMP8 inference against simultaneous mouse-neocortical electrophysiology.

FlyWire reconstructs chemical-synapse wiring in a sampled adult female Drosophila brain using electron microscopy and proofreading.

MICrONS combines visual-response calcium imaging and dense electron-microscopy reconstruction in one mouse visual-cortical volume.

Forward simulators encode reproduction, mutation, recombination, selection and population regulation through a specified life cycle.

Niche differences and near-neutral fitness similarities can act concurrently in a competition–colonization model.

The 2014 point/counterpoint compares emphasis on developmental bias, plasticity, niche construction and broader inheritance with the view that evolutionary theory already integrates such evidence.

Ecological change and heritable trait change can influence one another, but parallel time-series changes do not identify their causal directions.

An inverse solution is a posterior distribution conditioned on observations, a forward operator, likelihood and prior.

Linearizing d=g(m) about m0 gives a sensitivity matrix J.

Thermal radiance retrieval separates surface temperature from emissivity and atmospheric absorption/emission using a specified forward model and ancillary inputs.

Differential radar phase constrains displacement along the satellite line of sight after orbital, topographic and atmospheric corrections.

Ensemble assimilation estimates forecast-error covariances from model realizations and updates the state using observations and their error model.

Fingerprinting compares observed patterns with simulated responses to specified forcing and internal-variability covariance.

Dry atmospheric Ertel potential vorticity combines absolute vorticity and the potential-temperature gradient per unit density.

TEOS-10 seawater calculations distinguish Absolute Salinity, Conservative Temperature and sea pressure.

Mixing waters can change density through nonlinear temperature/salinity dependence even when each source has the same density at the reference pressure.

A mineral assemblage at specified pressure, temperature and bulk composition minimizes Gibbs energy subject to component balance.

Combining mineral phase fractions with elastic properties can predict seismic velocities, but equilibrium databases and shear-modulus data have separate calibration domains.

An apparent thermochronological age depends on diffusion kinetics, grain geometry and cooling history.

U–Pb discordance can reflect common lead, radiogenic lead loss, initial daughter disequilibrium or mixtures of differently aged growth zones.