ML3DClassification

ML3D Classification and Reconstruction

• Correct the grey-scale for the initial reference volume (only necessary for non-Xmipp volumes!)
• Low-pass filter the initial reference volume
• Random, unbiased seed generation
• ML3D classification

Critical parameters

• Click ML3DClassificationParameters for our recommendations on the most critical parameters of this protocol.

Example Data

• Input
  • gallery of images
  • reference volume

File ml3d_classification.tar.gz contains all the images, the reference volume and the xmipp_protocol_ml3d.py script necesary to run this protocol. This example consists 20,000 simulated ribosome images (64x64 pixels). The 75% of the images belong to a class without elongation factor G (EFG) and the rest belong to a class with EFG factor. The protocol is able to classify the images into the two different states.

• Output
  • Selection file per class
  • Log file with classification information
  • 3D reconstructions
  • Document file with optimal alignment parameters

File output_ml3d_classification.tar.gz contains the results of executing the[[ML3D]] script.

Pressanalyze results button for visualizing the results of the classification (3D reconstruction, classes, convergence rate, etc).

Reconstructed 3D maps without EFG (blue) and with EFG (pink)

NOTE: since this protocol uses random seeds, the results may vary between different runs.

USER's COMMENTS

--Main.RobertoMarabini - 09 Oct 2007