Intelligence

The Intelligence tab provides advanced features for specialized tasks.

Equilibrate

#equilibrate

Setup

#setup

Trajectory

#trajectory

Backmapping

#backmapping

Setup

#template-matching

Add

#add

Membrane

#membrane

Equilibrate

Prepares meshes for simulation:

1. Select a mesh model

2. Configure parameters:

   • Average Edge Length: Target edge length for mesh uniformity

   • Steps: Number of equilibration iterations

   • Energy coefficients: Control mesh equilibration

3. Choose directory: Click Equilibrate and select an output directory for simulation files

4. Click OK to start the equilibration

Output Files:

• mesh_base: Original input mesh

• mesh_remeshed: Mesh with target edge length

• mesh_equilibrated: Fully equilibrated mesh ready for simulation

• mesh.txt: Configuration file with mesh scale and offest parameters

Note

Requires Trimem for mesh equilibration (see installation instructions).

Setup

Configures Helfrich Monte Carlo Flexible Fitting (HMFF) simulation:

1. Click Setup and choose a directory containing equilibrated meshes.

2. Select from available equilibrated mesh files

3. Configure parameters:

   Input Data:

   • Volume: Select volume file with densities (MRC format)

   • Invert Contrast: Flip density values if needed

   HMFF Parameters:

   • HMFF Weight (ξ): Coupling strength to experimental data

   • Rigidity (κ): Membrane bending rigidity

   • Steps: Number of simulation steps

   • Threads: Parallel processing threads

   Filtering Options:

   • Lowpass Cutoff: High-frequency noise removal

   • Highpass Cutoff: Low-frequency artifact removal

4. Select output directory

5. Files are prepared for HMFF simulation

Simulation can be executed via the generated run.sh script.

Note

Requires FreeDTS for simulation (see installation instructions).

Trajectory

Imports and visualizes DTS simulation results:

1. Click the arrow next to Trajectory

2. Configure import settings:

   • Scale: Coordinate scaling factor (1/scale applied to points)

   • Offset: Coordinate offset (single value or x,y,z triplet)

3. Select directory with a DTS trajectory from FreeDTS

4. Trajectory is loaded into the Trajectory Player

Supported Formats:

• TSI: FreeDTS topology files (.tsi, .q)

• VTU: VTK unstructured grid files (.vtu)

Tip

Use View > Trajectory Player to navigate through time points.

Backmapping

Creates coarse-grained molecular models of membrane surfaces, optionally including protein positions and orientations.

1. Click Backmapping and choose location for generated coarse-grained files

2. elect a mesh model as the membrane surface

3. Configure parameters:

   Mesh Settings:

   • Target Edge Length: Spatial resolution for coarse-grained model

   Protein Inclusions:

   • Add mappings: Associate point cloud clusters with protein types

   • Include Normals: Preserve orientation information

   • Flip Normals: Reverse normal direction if needed

4. The system is prepared for multi-scale modeling

Output Files:

• mesh.tsi: DTS-compatible mesh with protein inclusions

• martinize.sh: Script for protein coarse-graining

• plm.sh: Bilayer generation script

• pcg.sh: Lipid population script

Note

Requires TS2CG for backmapping (see installation instructions).

Template Matching

Setup

Configures template matching for protein identification:

1. Click Setup in the Template Matching section

2. Configure data paths:

   • Input tomogram

   • Template structures

   • Output directory

3. Set matching parameters:

   • Angular sampling

   • Score function

   • Uncertainty values

4. Configure computational resources

5. Run template matching to identify protein positions

Segmentation Operations

Add

Creates new empty clusters for manual point addition:

1. Click Add in the Segmentation section

2. A new empty cluster appears in the Object Browser

3. Use drawing mode (A key) to manually add points

4. Or use for testing with random point generation

Note

This function is included for testing and might be removed in a future release.

Membrane Segmentation

Automatically segments cellular membranes in tomograms using MemBrain-seg:

1. Click Membrane

2. Select model path (neural network weights - typically a .cpt file)

3. Configure parameters:

   Model Settings:

   • Model Path: Location of MemBrain-seg checkpoint file

   • Window Size: Processing block size (160 recommended)

   • Augmentation: Enable test-time augmentation for robustness

   Post-processing:

   • Clustering: Group connected components

   • Sampling Rates: Input/output resolution scaling

4. Select tomogram file and run segmentation

The output will be automatically loaded into the GUI. A copy of the segmentation is created in $HOME/mosaic/segmentations/membrain.

Note

A GPU is required to perform membrane segmentation in reasonable time.