Job: Ab-Initio Reconstruction


  • Reconstruct a single (homogeneous) or multiple (heterogeneous) 3D maps from a set of particles, without any initial models or starting structures required.


  • Particles
  • Initial model (optional)

Common Parameters

  • Number of classes: The default number of classes is 1. Increase this value if you expect to find multiple conformations/states in your data, or if you are trying ab-initio reconstruction to identify junk particles.


  • 3D map(s)
  • Plots, including orientation distributions


  • Use ab-initio reconstruction to reconstruct one or more 3D maps from a set of particles, without any initial model required. You can also use this job to identify junk particles from the raw particle stack or following 2D classification.
  • If no initial map is provided, it is possible to discover 3-D classes that are significantly different. If an initial map is provided, it will be used as the starting structure for all classes.
  • When performing an ab-initio reconstruction on a symmetric structure, we recommend not enforcing symmetry.
  • For more information on ab-initio reconstruction, please see Nature Methods.


Ab-initio reconstruction can fail when data quality is poor or when viewing directions are missing or strongly biased. For strongly biased viewing directions, subsequent rounds of 2D classification and rebalancing of 2D classes may help equalize views. Highly symmetric structures can also pose challenges. It is not recommended to enforce symmetry during ab-initio reconstruction but can be helpful when symmetry is known in advance.

Common Next Steps

  • Homogeneous or Heterogeneous Refinement