This challenge aims to motivate the community to develop, benchmark, evaluate, and advance single-particle cryo-EM data processing workflows designed for data with conformational heterogeneity.

We provide a synthetic particle dataset containing conformations extracted from an unbiased atomistic molecular dynamics simulation, along with the sequence of the simulated protein. We challenge participants to determine an ensemble of atomic models that best represents the conformational dynamics of the protein.

We seek to motivate the community to develop new techniques and/or adapt existing workflows to reconstruct ensembles of atomic models from cryo-EM data containing continuous conformational heterogeneity.

Participants should attempt to:

• Recover the atomic conformations present in the cryo-EM dataset.
• Recover the full range of conformational heterogeneity present in the cryo-EM dataset.

Dataset

<aside> <img src="/icons/arrow-down-line_gray.svg" alt="/icons/arrow-down-line_gray.svg" width="40px" />

Particle .mrcs stacks, an associated .star file, and a FASTA sequence can be downloaded via Globus here (~240 GB).

</aside>

This dataset contains particles representing a single protein trajectory and corresponding amino acid sequence.

• 1,753,470 particles, 1.0 Å/pixel, 256 pixel box size
• Consensus reconstruction poses are provided; particles are not centered
• Per-particle CTF defocus values are provided
• Detector MTF is provided if helpful

Evaluation

Entries will be evaluated by the accuracy of the atomic models, how similar the local heterogeneity is to the ground truth, and whether the full conformational change is recovered.

Metrics for this assessment are summarized here:

Scoring Metrics

Submission

<aside> <img src="/icons/folder_gray.svg" alt="/icons/folder_gray.svg" width="40px" />

Entries can be uploaded and submitted via Google Forms here.

</aside>

Please provide:

• Workflow summary

• Between 10 and 100 atomic models (mmCIF format) representing the conformations present in the dataset

  <aside> <img src="/icons/warning_gray.svg" alt="/icons/warning_gray.svg" width="40px" />

  Each model must be deposited as a separate mmCIF file.

  All models must be aligned.

  Each model must be complete as follows:

  • Whole chain defined with complete residues; no partial side chains
  • All expected heavy atoms included (from provided sequence)
  • All hydrogen atoms removed

  All parameters of the submitted models will be considered in evaluation, except for atomic B-factors and occupancies.

  Each model will be weighted equally when calculating evaluation metrics.

  </aside>

Optionally (if your method provides them), please provide:

• Latent space representation of the dataset (.csv file):

  Metadata Format: Latent Space

• Density map representations corresponding to particles (.csv file and .mrc files):

  Metadata Format: Density Maps

  <aside> <img src="/icons/warning_gray.svg" alt="/icons/warning_gray.svg" width="40px" />

  These representations may be used to create a community dataset for metrics development but will not be assessed.

  </aside>