Crafting a Lower CC Lens for Reducing Cost and Specimen Damage in Cryo-EM
Bibliographic record
Abstract
Single particle cryo-EM is poised to surpass x-ray crystallography as the most successful technique for resolving membrane protein structures, but the standard 300kV acceleration voltage of most commercially available cryomicroscopes contributes heavily to radiation damage and is highly expensive to maintain [1]. Recently published work by Russo et al., from the Medical Research Council Laboratory of Molecular Biology (MRC-LMB) in Cambridge, has advocated for a low (<2mm) chromatic aberration coefficient (CC) objective lens (OL) for use in a purpose-built 100kV cryo-TEM [1], which they determine to be the optimal electron energy for structural determination of many biological specimens in terms of maximizing information at minimal cost [2-4]. This previously unmet target specification requires a bespoke OL pole-piece designed to minimize CC. This necessitates the testing of a large multitude of candidate geometries in accordance with various parameters, which is greatly time consuming if done manually. By writing an evolutionary algorithm that defines and evaluates many geometries in parallel, using multiphysics and electron optics simulations, we have achieved the accelerated design and manufacture of such a pole-piece. The algorithm operates using the Dawninan theory of evolution, by creating an initial generation of candidates, evaluating their fitness with a customized cost function, then constructing a new generation using elitism (keeping the best-performing candidates from the previous generation), crossover (creating offspring geometries which mix the characteristics of two existing candidates), and mutation (randomly altering one or more nodes of an existing geometry). For each candidate in a generation, magnetic flux simulations are performed by COMSOL Multiphysics, which outputs comma separated values of flux data along the optical axis. This data is utilized by an electron optics simulation written in Julia, which calculates the lens aberrations and the back focal plane. Python then evaluates the candidate in accordance with the fitness function and ranks it against the rest of the candidates in that generation. It thereby categorizes the candidates into elites, mutatables and crossover parents, and accordingly creates the next generation of candidates to be tested. This is repeated until the generation quota is reached. A visual summary of the algorithmic process is displayed in Fig. 1. A dashboard of plots, generated for the fittest member of each generation, is shown in Fig. 2 for one sample candidate geometry. The algorithm generated a machinable pole-piece design with a simultaed CC nearing sub-1mm, a promising result. From this, a physical prototype of the final candidate geometry was developed and installed in a JEOL 1400 model TEM. It is currently undergoing testing by the Russo group in MRC-LMB, with results (experimental CC value and microscope images) expected by early summer. Future work will include improving the parallelization of the multiphysics simulations for increased efficiency, extending the functionality to more TEM models, and designing and evaluating new cost functions to accommodate for a wider range of applications (e.g. tomography, spectroscopy, in-situ microscopy etc.) [5]. (left) Illustration of the process used by the algorithm to create a new generation of candidates. The user can specify the proportions in which the generations are categorized. (right) flow chart outlining the algorithmic process. Dashboard for monitoring the geometry evolution. Main figure shows a cross-section of the current geometry, subplots show the candidate's magnetic flux profile and the tracking of various parameters, most importantly aberration coefficients, as the number of generations increases.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.027 | 0.006 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".