pith. sign in

arxiv: 2410.13713 · v2 · submitted 2024-10-17 · 💻 cs.LG

CrystalX: High-accuracy Crystal Structure Analysis Using Deep Learning

Kaipeng Zheng , Weiran Huang , Wanli Ouyang , Han-Sen Zhong , Yuqiang Li This is my paper

Pith reviewed 2026-05-23 18:35 UTC · model grok-4.3

classification 💻 cs.LG
keywords crystal structure analysisdeep learningX-ray diffractionautomated structure determinationcrystallography automationstructure validation
0
0 comments X

The pith

CrystalX uses deep learning to automate full-atom crystal structure analysis from X-ray diffraction data and outperforms prior automated methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents CrystalX as a deep learning approach to fully automated crystal structure determination at the atomic level from experimental X-ray measurements. Training and testing rely on more than 50,000 real diffraction datasets split strictly by publication date so that test structures come from later papers than those seen in training. The model exceeds the performance of existing automated baselines, identifies geometric patterns in complex cases, and corrects interpretation mistakes in peer-reviewed structures that pass standard CheckCIF checks. These capabilities allow the method to handle newly discovered compounds without human input in routine laboratory pipelines.

Core claim

CrystalX, a deep learning model, performs high-accuracy automated structure analysis of crystalline materials from X-ray diffraction data at the full-atom level. With training and test sets separated by publication time, the model surpasses automated baselines, deciphers intricate patterns, and rectifies expert errors in published structures that evade CheckCIF A/B alerts, supporting its deployment for human-free analysis of new compounds.

What carries the argument

The CrystalX deep learning model trained on X-ray diffraction measurements to predict full atomic crystal structures.

If this is right

  • Routine crystal structure analysis can proceed without human intervention for compounds discovered after the training period.
  • Errors in published structures that escape automated validation alerts can be identified and corrected by the model.
  • Self-driving laboratories gain the capacity for end-to-end automated structural characterization.
  • The approach scales to daily processing of new experimental diffraction measurements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Integration with robotic synthesis systems could shorten the cycle from material discovery to confirmed structure.
  • Similar models might be trained on other diffraction or spectroscopic modalities for broader material identification tasks.
  • Widespread adoption could shift the bottleneck in crystallography from data interpretation to data collection.
  • The method provides a benchmark for measuring how much expert judgment remains necessary after automated analysis.

Load-bearing premise

The temporal split of data by publication date suffices to show that the model will correctly analyze structures of compounds never seen during training.

What would settle it

Apply CrystalX to a collection of crystal structures published after the training cutoff and compare its output atomic positions and space groups against independent expert manual determinations on the same data.

read the original abstract

Atomic structure analysis of crystalline materials is a paramount endeavor in both chemical and material sciences. This sophisticated technique necessitates not only a solid foundation in crystallography but also a profound comprehension of the intricacies of the accompanying software, posing a significant challenge in meeting the rigorous daily demands. For the first time, we confront this challenge head-on by harnessing the power of deep learning for fully automated routine structure analysis at the full-atom level. To validate the performance of the model, named CrystalX, we employed a dataset comprising over 50,000 X-ray diffraction measurements derived from authentic experiments. Under a strict temporal validation scheme that separates training and test data by publication time, CrystalX substantially outperformed the automated baseline and adept at deciphering intricate geometric patterns. Remarkably, CrystalX revealed that even peer-reviewed publications harbor expert interpretation errors that can evade stringent CheckCIF A/B-level alerts, yet CrystalX adeptly rectifies them. It has already been successfully applied in our day-to-day pipeline, enabling fully automated, human-free structure analysis for newly discovered compounds. Overall, CrystalX marks the beginning of a new era in automating routine structural analysis within self-driving laboratories.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces CrystalX, a deep learning model for fully automated, full-atom crystal structure analysis from X-ray diffraction measurements. It reports results on a dataset of over 50,000 experimental structures using a strict temporal split (training and test separated by publication date), claims substantial outperformance over automated baselines, the ability to detect and correct expert interpretation errors missed by CheckCIF A/B alerts, and successful deployment in a day-to-day pipeline for newly discovered compounds.

Significance. If the performance and generalization claims hold with rigorous supporting evidence, the work could enable substantial automation of routine crystallography, reducing expert time in self-driving laboratories and potentially improving data quality by catching subtle errors. The temporal-split validation approach is a positive step toward realistic evaluation, though its sufficiency for true novelty remains to be demonstrated.

major comments (2)
  1. [Abstract] Abstract: The central claim that CrystalX enables reliable human-free analysis for newly discovered compounds rests on the temporal validation demonstrating out-of-distribution performance. However, no analysis is provided showing that test-set structures differ from the training distribution in chemical composition, space-group statistics, or geometric motifs; publication-date separation alone permits substantial overlap, so outperformance on the held-out set does not establish generalization to genuinely novel compounds.
  2. [Abstract] Abstract: No model architecture, training procedure, quantitative metrics (e.g., R-factors, success rates, error distributions), or error analysis are reported, preventing evaluation of whether the stated performance advantage over the automated baseline is statistically or practically meaningful.
minor comments (1)
  1. [Abstract] Abstract: The sentence 'CrystalX substantially outperformed the automated baseline and adept at deciphering intricate geometric patterns' is grammatically incomplete.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their comments. We address each major comment below, indicating planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that CrystalX enables reliable human-free analysis for newly discovered compounds rests on the temporal validation demonstrating out-of-distribution performance. However, no analysis is provided showing that test-set structures differ from the training distribution in chemical composition, space-group statistics, or geometric motifs; publication-date separation alone permits substantial overlap, so outperformance on the held-out set does not establish generalization to genuinely novel compounds.

    Authors: We agree that publication-date separation alone does not guarantee distributional shift across all relevant features. In the revised manuscript we will add a direct comparison of the train and test sets on chemical composition, space-group frequencies, and representative geometric motifs to better substantiate the out-of-distribution character of the test data. revision: yes

  2. Referee: [Abstract] Abstract: No model architecture, training procedure, quantitative metrics (e.g., R-factors, success rates, error distributions), or error analysis are reported, preventing evaluation of whether the stated performance advantage over the automated baseline is statistically or practically meaningful.

    Authors: The abstract is intentionally concise. The full manuscript contains the model architecture, training details, quantitative metrics (including success rates and R-factor comparisons), and error analysis. To make the abstract more self-contained we will insert the principal quantitative performance figures into the revised abstract. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical ML evaluation on external temporal data split

full rationale

The paper presents an applied deep-learning system for crystal structure determination with no equations, derivations, or first-principles claims. Performance is assessed via a temporal train/test split on >50k real experimental X-ray datasets, which is an independent external benchmark rather than any quantity fitted from or defined by the target result. No self-citations, ansatzes, or uniqueness theorems are invoked to justify the core method; the reported outperformance and error-correction examples rest on direct comparison to baselines and CheckCIF on held-out data. This is the standard non-circular pattern for supervised ML papers whose central claim is empirical generalization on real measurements.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on a deep neural network whose parameters are fitted to the 50k experimental dataset and on the domain assumption that publication-date separation guarantees generalization to future compounds.

free parameters (1)
  • Neural network parameters
    Weights and biases of the deep learning model are optimized on the training portion of the 50,000-measurement dataset.
axioms (1)
  • domain assumption Publication-date temporal split prevents information leakage and tests generalization to newly discovered compounds
    Invoked to justify the validation scheme described in the abstract.

pith-pipeline@v0.9.0 · 5742 in / 1225 out tokens · 58548 ms · 2026-05-23T18:35:41.083705+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

51 extracted references · 51 canonical work pages

  1. [1]

    Nature 171(4356), 737–738 (1953)

    Watson, J.D., Crick, F.H.: Molecular structure of nucleic acids: a structure for deoxyribose nucleic acid. Nature 171(4356), 737–738 (1953)

    work page 1953

  2. [2]

    Chemistry of penicillin, 310–367 (1949)

    Crowfoot, D., Bunn, C.W., Rogers-Low, B.W., Turner-Jones, A.: X-ray crys- tallographic investigation of the structure of penicillin. Chemistry of penicillin, 310–367 (1949)

    work page 1949

  3. [3]

    Nature 176(4477), 325–328 (1955)

    DG, H., Pickworth, J., JH, R., KN, T., RJ, P., JG, W.: The crystal structure of the hexacarboxylic acid derived from b12 and the molecular structure of the vitamin. Nature 176(4477), 325–328 (1955)

    work page 1955

  4. [4]

    Nature 224(5218), 491–495 (1969)

    Adams, M.J., Blundell, T.L., Dodson, E.J., Dodson, G.G., Vijayan, M., Baker, E.N., Harding, M.M., Hodgkin, D.C., Rimmer, B., Sheat, S.: Structure of rhombohedral 2 zinc insulin crystals. Nature 224(5218), 491–495 (1969)

    work page 1969

  5. [5]

    Science 347(6227), 1221–1226 (2015)

    Li, J., Ballmer, S.G., Gillis, E.P., Fujii, S., Schmidt, M.J., Palazzolo, A.M., Lehmann, J.W., Morehouse, G.F., Burke, M.D.: Synthesis of many different types 10 of organic small molecules using one automated process. Science 347(6227), 1221–1226 (2015)

    work page 2015

  6. [6]

    Angewandte Chemie International Edition 57(16), 4192–4214 (2018)

    Trobe, M., Burke, M.D.: The molecular industrial revolution: automated synthesis of small molecules. Angewandte Chemie International Edition 57(16), 4192–4214 (2018)

    work page 2018

  7. [7]

    Nature communications 9(1), 2849 (2018)

    Greenaway, R., Santolini, V., Bennison, M., Alston, B., Pugh, C., Little, M., Mik- litz, M., Eden-Rump, E., Clowes, R., Shakil, A.,et al.: High-throughput discovery of organic cages and catenanes using computational screening fused with robotic synthesis. Nature communications 9(1), 2849 (2018)

    work page 2018

  8. [8]

    : Automated iterative c sp 3–c bond formation

    Blair, D.J., Chitti, S., Trobe, M., Kostyra, D.M., Haley, H.M., Hansen, R.L., Ballmer, S.G., Woods, T.J., Wang, W., Mubayi, V., et al. : Automated iterative c sp 3–c bond formation. Nature 604(7904), 92–97 (2022)

    work page 2022

  9. [9]

    Nature Synthesis, 1–8 (2024)

    Wang, W., Angello, N.H., Blair, D.J., Tyrikos-Ergas, T., Krueger, W.H., Medine, K.N., LaPorte, A.J., Berger, J.M., Burke, M.D.: Rapid automated iterative small- molecule synthesis. Nature Synthesis, 1–8 (2024)

    work page 2024

  10. [10]

    Metherall, J.P., Carroll, R.C., Coles, S.J., Hall, M.J., Probert, M.R.: Advanced crystallisation methods for small organic molecules. Chem. Soc. Rev. 52, 1995– 2010 (2023)

    work page 1995

  11. [11]

    Chemical reviews 122(13), 11514–11603 (2022)

    Bolla, G., Sarma, B., Nangia, A.K.: Crystal engineering of pharmaceutical cocrys- tals in the discovery and development of improved drugs. Chemical reviews 122(13), 11514–11603 (2022)

    work page 2022

  12. [12]

    Chem 6(7), 1755–1765 (2020)

    Tyler, A.R., Ragbirsingh, R., McMonagle, C.J., Waddell, P.G., Heaps, S.E., Steed, J.W., Thaw, P., Hall, M.J., Probert, M.R.: Encapsulated nanodroplet crystallization of organic-soluble small molecules. Chem 6(7), 1755–1765 (2020)

    work page 2020

  13. [13]

    Acta Crystallographica Section A: Foundations and Advances 71(1), 3–8 (2015)

    Sheldrick, G.M.: Shelxt–integrated space-group and crystal-structure determina- tion. Acta Crystallographica Section A: Foundations and Advances 71(1), 3–8 (2015)

    work page 2015

  14. [14]

    Acta Crystallographica Section A: Foundations of Crystallography 64(1), 112–122 (2008)

    Sheldrick, G.M.: A short history of shelx. Acta Crystallographica Section A: Foundations of Crystallography 64(1), 112–122 (2008)

    work page 2008

  15. [15]

    Acta Crystallographica Section A: Foundations of Crystallography 60(2), 134–141 (2004)

    Oszl´ anyi, G., S¨ ut˝ o, A.: Ab initio structure solution by charge flipping. Acta Crystallographica Section A: Foundations of Crystallography 60(2), 134–141 (2004)

    work page 2004

  16. [16]

    Journal of applied crystallogra- phy 32(1), 115–119 (1999) 11

    Altomare, A., Burla, M.C., Camalli, M., Cascarano, G.L., Giacovazzo, C., Guagliardi, A., Moliterni, A.G., Polidori, G., Spagna, R.: Sir97: a new tool for crystal structure determination and refinement. Journal of applied crystallogra- phy 32(1), 115–119 (1999) 11

    work page 1999

  17. [17]

    Journal of Applied Crystallography 48(1), 306–309 (2015)

    Burla, M.C., Caliandro, R., Carrozzini, B., Cascarano, G.L., Cuocci, C., Giacov- azzo, C., Mallamo, M., Mazzone, A., Polidori, G.: Crystal structure determination and refinement via SIR2014. Journal of Applied Crystallography 48(1), 306–309 (2015)

    work page 2015

  18. [18]

    Science 385(6708), 522–528 (2024)

    Larsen, A.S., Rekis, T., Madsen, A.Ø.: Phai: A deep-learning approach to solve the crystallographic phase problem. Science 385(6708), 522–528 (2024)

    work page 2024

  19. [19]

    Acta Crystallographica Section C: Structural Chemistry 71(1), 3–8 (2015)

    Sheldrick, G.M.: Crystal structure refinement with shelxl. Acta Crystallographica Section C: Structural Chemistry 71(1), 3–8 (2015)

    work page 2015

  20. [20]

    Journal of Applied Crystallography 36(6), 1487–1487 (2003)

    Betteridge, P.W., Carruthers, J.R., Cooper, R.I., Prout, K., Watkin, D.J.: Crys- tals version 12: software for guided crystal structure analysis. Journal of Applied Crystallography 36(6), 1487–1487 (2003)

    work page 2003

  21. [21]

    Journal of applied crystallography 42(2), 339–341 (2009)

    Dolomanov, O.V., Bourhis, L.J., Gildea, R.J., Howard, J.A., Puschmann, H.: Olex2: a complete structure solution, refinement and analysis program. Journal of applied crystallography 42(2), 339–341 (2009)

    work page 2009

  22. [22]

    Journal of applied crystallography 44(6), 1281–1284 (2011)

    H¨ ubschle, C.B., Sheldrick, G.M., Dittrich, B.: Shelxle: a qt graphical user interface for shelxl. Journal of applied crystallography 44(6), 1281–1284 (2011)

    work page 2011

  23. [23]

    journal of Applied Crystallography 32(4), 837–838 (1999)

    Farrugia, L.J.: Wingx suite for small-molecule single-crystal crystallography. journal of Applied Crystallography 32(4), 837–838 (1999)

    work page 1999

  24. [24]

    nature 542(7639), 115–118 (2017)

    Esteva, A., Kuprel, B., Novoa, R.A., Ko, J., Swetter, S.M., Blau, H.M., Thrun, S.: Dermatologist-level classification of skin cancer with deep neural networks. nature 542(7639), 115–118 (2017)

    work page 2017

  25. [25]

    Nature 624(7990), 80–85 (2023)

    Merchant, A., Batzner, S., Schoenholz, S.S., Aykol, M., Cheon, G., Cubuk, E.D.: Scaling deep learning for materials discovery. Nature 624(7990), 80–85 (2023)

    work page 2023

  26. [26]

    : A principal odor map unifies diverse tasks in olfactory perception

    Lee, B.K., Mayhew, E.J., Sanchez-Lengeling, B., Wei, J.N., Qian, W.W., Little, K.A., Andres, M., Nguyen, B.B., Moloy, T., Yasonik, J., et al. : A principal odor map unifies diverse tasks in olfactory perception. Science 381(6661), 999–1006 (2023)

    work page 2023

  27. [27]

    : Improved protein structure prediction using potentials from deep learning

    Senior, A.W., Evans, R., Jumper, J., Kirkpatrick, J., Sifre, L., Green, T., Qin, C., ˇZ´ ıdek, A., Nelson, A.W., Bridgland, A., et al. : Improved protein structure prediction using potentials from deep learning. Nature577(7792), 706–710 (2020)

    work page 2020

  28. [28]

    Nature Methods 19(6), 730–739 (2022)

    Tubiana, J., Schneidman-Duhovny, D., Wolfson, H.J.: Scannet: an interpretable geometric deep learning model for structure-based protein binding site prediction. Nature Methods 19(6), 730–739 (2022)

    work page 2022

  29. [29]

    12 IUCrJ 4(4), 486–494 (2017)

    Park, W.B., Chung, J., Jung, J., Sohn, K., Singh, S.P., Pyo, M., Shin, N., Sohn, K.-S.: Classification of crystal structure using a convolutional neural network. 12 IUCrJ 4(4), 486–494 (2017)

    work page 2017

  30. [30]

    Nature communications 9(1), 2775 (2018)

    Ziletti, A., Kumar, D., Scheffler, M., Ghiringhelli, L.M.: Insightful classification of crystal structures using deep learning. Nature communications 9(1), 2775 (2018)

    work page 2018

  31. [31]

    npj Computational Materials 5(1), 60 (2019)

    Oviedo, F., Ren, Z., Sun, S., Settens, C., Liu, Z., Hartono, N.T.P., Ramasamy, S., DeCost, B.L., Tian, S.I., Romano, G., et al.: Fast and interpretable classifica- tion of small x-ray diffraction datasets using data augmentation and deep neural networks. npj Computational Materials 5(1), 60 (2019)

    work page 2019

  32. [32]

    Science 367(6477), 564–568 (2020)

    Kaufmann, K., Zhu, C., Rosengarten, A.S., Maryanovsky, D., Harrington, T.J., Marin, E., Vecchio, K.S.: Crystal symmetry determination in electron diffraction using machine learning. Science 367(6477), 564–568 (2020)

    work page 2020

  33. [33]

    Journal of applied crystallography 42(4), 726–729 (2009)

    Graˇ zulis, S., Chateigner, D., Downs, R.T., Yokochi, A., Quir´ os, M., Lutterotti, L., Manakova, E., Butkus, J., Moeck, P., Le Bail, A.: Crystallography open database– an open-access collection of crystal structures. Journal of applied crystallography 42(4), 726–729 (2009)

    work page 2009

  34. [34]

    https://clarivate.com/products/scientific-and-academic-research/research- analytics-evaluation-and-management-solutions/journal-citation-reports/

    work page

  35. [35]

    Advances in neural information processing systems 30 (2017)

    Sch¨ utt, K., Kindermans, P.-J., Sauceda Felix, H.E., Chmiela, S., Tkatchenko, A., M¨ uller, K.-R.: Schnet: A continuous-filter convolutional neural network for mod- eling quantum interactions. Advances in neural information processing systems 30 (2017)

    work page 2017

  36. [36]

    In: International Conference on Learning Representations (2020)

    Gasteiger, J., Groß, J., G¨ unnemann, S.: Directional message passing for molecular graphs. In: International Conference on Learning Representations (2020)

    work page 2020

  37. [37]

    In: International Conference on Learning Representations (2022)

    Liu, Y., Wang, L., Liu, M., Lin, Y., Zhang, X., Oztekin, B., Ji, S.: Spherical message passing for 3d molecular graphs. In: International Conference on Learning Representations (2022)

    work page 2022

  38. [38]

    In: Oh, A.H., Agarwal, A., Belgrave, D., Cho, K

    Wang, L., Liu, Y., Lin, Y., Liu, H., Ji, S.: ComENet: Towards complete and efficient message passing for 3d molecular graphs. In: Oh, A.H., Agarwal, A., Belgrave, D., Cho, K. (eds.) Advances in Neural Information Processing Systems (2022)

    work page 2022

  39. [39]

    In: International Conference on Learning Representations (2022)

    Th¨ olke, P., Fabritiis, G.D.: Equivariant transformers for neural network based molecular potentials. In: International Conference on Learning Representations (2022)

    work page 2022

  40. [40]

    Science 355(6321), 166–169 (2017) 13

    Palatinus, L., Br´ azda, P., Boullay, P., Perez, O., Klementov´ a, M., Petit, S., Eigner, V., Zaarour, M., Mintova, S.: Hydrogen positions in single nanocrystals revealed by electron diffraction. Science 355(6321), 166–169 (2017) 13

    work page 2017

  41. [41]

    Acta Crystallographica Section A: Foundations and Advances 75(1), 82–93 (2019)

    Clabbers, M.T., Gruene, T., Genderen, E., Abrahams, J.P.: Reducing dynamical electron scattering reveals hydrogen atoms. Acta Crystallographica Section A: Foundations and Advances 75(1), 82–93 (2019)

    work page 2019

  42. [42]

    Acta Crystallo- graphica Section D: Biological Crystallography 65(2), 148–155 (2009)

    Spek, A.L.: Structure validation in chemical crystallography. Acta Crystallo- graphica Section D: Biological Crystallography 65(2), 148–155 (2009)

    work page 2009

  43. [43]

    In: Juraf- sky, D., Chai, J., Schluter, N., Tetreault, J

    Abnar, S., Zuidema, W.: Quantifying attention flow in transformers. In: Juraf- sky, D., Chai, J., Schluter, N., Tetreault, J. (eds.) Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 4190–4197. Association for Computational Linguistics, Online (2020)

    work page 2020

  44. [44]

    Journal of machine learning research 9(11) (2008)

    Maaten, L., Hinton, G.: Visualizing data using t-sne. Journal of machine learning research 9(11) (2008)

    work page 2008

  45. [45]

    Nature Synthesis 2(6), 483–492 (2023)

    Abolhasani, M., Kumacheva, E.: The rise of self-driving labs in chemical and materials sciences. Nature Synthesis 2(6), 483–492 (2023)

    work page 2023

  46. [46]

    Nature Chemistry 15(6), 848–855 (2023)

    Klar, P.B., Krysiak, Y., Xu, H., Steciuk, G., Cho, J., Zou, X., Palatinus, L.: Accu- rate structure models and absolute configuration determination using dynamical effects in continuous-rotation 3d electron diffraction data. Nature Chemistry 15(6), 848–855 (2023)

    work page 2023

  47. [47]

    Journal of the American Chemical Society 143(23), 8713–8719 (2021)

    Kapaca, E., Jiang, J., Cho, J., Jord´ a, J.L., Diaz-Cabanas, M.J., Zou, X., Corma, A., Willhammar, T.: Synthesis and structure of a 22 × 12× 12 extra-large pore zeolite itq-56 determined by 3d electron diffraction. Journal of the American Chemical Society 143(23), 8713–8719 (2021)

    work page 2021

  48. [48]

    Chemical Reviews 122(17), 13883–13914 (2022)

    Saha, A., Nia, S.S., Rodr´ ıguez, J.A.: Electron diffraction of 3d molecular crystals. Chemical Reviews 122(17), 13883–13914 (2022)

    work page 2022

  49. [49]

    Nature 601(7893), 360–365 (2022)

    Schriber, E.A., Paley, D.W., Bolotovsky, R., Rosenberg, D.J., Sierra, R.G., Aquila, A., Mendez, D., Poitevin, F., Blaschke, J.P., Bhowmick, A., et al.: Chem- ical crystallography by serial femtosecond x-ray diffraction. Nature 601(7893), 360–365 (2022)

    work page 2022

  50. [50]

    Nature chemistry 15(4), 491–497 (2023)

    Takaba, K., Maki-Yonekura, S., Inoue, I., Tono, K., Hamaguchi, T., Kawakami, K., Naitow, H., Ishikawa, T., Yabashi, M., Yonekura, K.: Structural resolu- tion of a small organic molecule by serial x-ray free-electron laser and electron crystallography. Nature chemistry 15(4), 491–497 (2023)

    work page 2023

  51. [51]

    A web application is available at https://crystalx.intern-ai.org.cn/, enabling users to upload their own data

    https://doi.org/10.5281/zenodo.13820303 Data and code availability: All data required to validate the conclusions of this paper are provided in the manuscript, the supplementary materials, and a data archive containing all relevant 14 data, code, and model parameters, accessible at [51]. A web application is available at https://crystalx.intern-ai.org.cn/...

    work page doi:10.5281/zenodo.13820303