Molecular Cell's AI Policy: Cell Press Rules for Structural and Molecular Biology Authors

This comes up constantly for Molecular Cell authors. Does showing an AlphaFold-predicted structure as a figure violate the AI-generated image ban?

No. Here's why:

AlphaFold (and similar tools like ESMFold, RoseTTAFold, OmegaFold) generates computational predictions from protein sequence data. The output is a structural model, a scientific result derived from data through a defined computational method. This is fundamentally different from asking Midjourney to "draw a protein structure" based on a text prompt.

The same principle applies to:

• Molecular dynamics simulation snapshots, computational results, not generated images
• Docking predictions rendered as figures, scientific outputs
• Homology models built with SWISS-MODEL or similar, computational results
• Electrostatic surface maps calculated from structure data, data-derived visualizations

What IS prohibited:

• Using DALL-E to generate a schematic of protein-protein interactions
• Using Midjourney to create a graphical abstract showing molecular mechanisms
• Using any generative AI to illustrate a biological concept that isn't derived from data

How to describe AlphaFold in STAR Methods:

AlphaFold belongs in your standard Method Details as a computational tool:

"Protein structure predictions were generated using AlphaFold2 (Jumper et al., 2021) with default parameters. Predicted structures with pLDDT scores >70 were used for subsequent analysis. Structural figures were rendered in PyMOL (Schrödinger)."

This is research methodology, not an AI writing disclosure. It goes in the computational methods subsection, not alongside your ChatGPT disclosure.

For a structural biology paper:

"Protein structure predictions were performed using AlphaFold2 as described in STAR Methods: Structural Prediction. Separately, during manuscript preparation, the authors used ChatGPT (GPT-4, OpenAI) to improve the language clarity of the Discussion section. All AI-suggested text edits were reviewed by the corresponding author (J.K.). The authors take full responsibility for the published content."

For a paper with cryo-EM data:

"Cryo-EM data processing was performed using cryoSPARC v4 and RELION 4.0 (see STAR Methods: Cryo-EM Data Processing). During manuscript preparation, GitHub Copilot (Microsoft) was used to assist with writing Python scripts for automated particle picking parameter optimization. ChatGPT (GPT-4, OpenAI) was used to improve the readability of the Results section. All code was validated against established processing workflows, and all text edits were reviewed by the authors."

For a biochemistry/enzymology paper:

"During preparation of this manuscript, the authors used Claude (Claude 3.5, Anthropic) to edit the Introduction for language clarity and to assist with formatting the kinetic parameters table. All content was verified against the experimental data by the senior author (M.R.). The authors take full responsibility for the content of this article."

Molecular Cell publishes many cryo-EM structures. The software stack for cryo-EM processing is extensive:

Research tools (STAR Methods, standard description): RELION, cryoSPARC, MotionCor2, CTFFIND, Phenix, Coot, ChimeraX

AI-assisted tools that are research methods: DeepEMhancer (AI-based map sharpening), Topaz (AI-based particle picking), ModelAngelo (AI-based model building)

Writing tools requiring disclosure: ChatGPT for editing text, Copilot for writing processing scripts

The middle category, AI tools used for cryo-EM processing, belongs in your standard STAR Methods as part of your research pipeline. These are computational tools, not manuscript preparation aids. Describe them the way you'd describe RELION: name, version, parameters, what they were used for.

Cell Press follows standard COPE-guided enforcement:

During review:

• Request to add disclosure to STAR Methods
• If AI involvement in model building or structure validation is suspected, additional review may be required
• Deliberate concealment can lead to rejection

After publication:

• Correction for undisclosed language editing
• Expression of concern if AI affected structural analysis or interpretation
• Retraction for fabricated data or false structural claims

The structural biology stakes: If you report a cryo-EM structure or crystal structure in Molecular Cell and didn't disclose AI involvement in the model building or refinement process, the structure's credibility is at stake. PDB depositions are permanent scientific records. If a deposited structure is later found to have undisclosed AI involvement in its building, this creates problems for everyone who cited or used that structure.

To be clear: using AI tools like ModelAngelo or DeepEMhancer for structure determination is perfectly legitimate and should be described in Methods as computational tools. The issue is when AI use in any part of the process, including writing, isn't properly documented.

Nature Structural & Molecular Biology (NSMB) is Molecular Cell's most direct competitor. NSMB follows Springer Nature's AI policy with free-form Methods disclosure; Molecular Cell uses STAR Methods. The substantive requirements are identical. If you're preparing a manuscript for both journals as backup options, the AI disclosure content will be the same, only the formatting changes.

For structural biology papers:

• Clearly distinguish AlphaFold/computational predictions (research tools) from ChatGPT/Copilot (writing tools) in STAR Methods
• If you used AI-based tools for model building (ModelAngelo), describe them alongside RELION and Phenix in your processing pipeline
• Structural figures rendered in PyMOL or ChimeraX from real or predicted structures aren't AI-generated images

For biochemistry and biophysics papers:

• If AI helped with kinetic modeling code, disclose this and confirm the models were validated against raw data
• For binding assay analysis, specify whether AI assisted with curve fitting code
• Don't use AI to generate hypothetical binding models in the Discussion unless clearly labeled as speculative

For gene regulation and epigenetics papers:

• If AI helped with ChIP-seq, ATAC-seq, or Hi-C analysis code, disclose and validate
• Genomic data processing tools (BWA, MACS2, HiC-Pro) are research tools in standard STAR Methods
• Custom scripts written with AI assistance need separate disclosure

Before submission checklist:

• [ ] AI disclosure in STAR Methods → Method Details
• [ ] Research tools (AlphaFold, RELION, cryoSPARC) in standard STAR Methods
• [ ] Writing tools (ChatGPT, Copilot) in separate AI disclosure
• [ ] No generative AI images or graphical abstract
• [ ] Structural figures derived from data (PyMOL, ChimeraX renders are fine)
• [ ] All code validated and deposited
• [ ] PDB deposition includes accurate method descriptions
• [ ] All co-authors reviewed AI disclosure

A Molecular Cell submission readiness check can help confirm your Molecular Cell submission meets Cell Press requirements before submission.

Comply proactively if:

• You used any AI tool (ChatGPT, Grammarly, Copilot) during manuscript preparation
• The journal requires AI use disclosure in the methods or acknowledgments
• Your institution has its own AI use policy that may be stricter

Less concerned if:

• You used AI only for grammar/spell checking (most journals exempt this)
• The journal does not have a formal AI policy yet
• Your use was limited to literature search or reference management