Skip to content

3D Functions

oluwatosin oluwadare edited this page Feb 19, 2019 · 16 revisions

3D model reconstruction by LorDG

A. Purpose
To build 3D chromosomes and genome models

B. Input
A contact matrix in sparse matrix format

C. Output
3D models in .gss format file and .pdb format file

D. Running Access the function from the menu toolbar: 3D-Functions/LorDG-3D Modeler

E. LorDG GUI

Field Description Default
Conversion Factor α in the formula d(i,j)=1/(IF(ij)^α), where IF(i,j) is interaction frequency between i and j. When the field is left blank, the program will search for the best value in the range [0.1-3.0]with a step size of 0.1. Users can also specify a range to search for by putting two numbers separated by a hyphen (e.g. 0.5-1.0). During the searching, the right-top corner of the main screen displays information about the current value being tested. 0.5
Initial Learning Rate Initial learning rate of the optimization. Higher learning rate can speed up the reconstruction process but can cause the process to fail as well 1.0
Max Number of Iterations Maximum number of iterations for the optimization 2000
Chromosome Chromosome name of the contact matrix in the input. If the input contains contact matrix of the whole genome, leave this field blank. X
Genome ID Genome version of the contact matrix in the input. hg19
Is Multiple-Chromosomes Structure? if the input contains both inter-and intra-chromosomal contacts data, this checkbox should be checked. unchecked
Length of Chromosomes This field contains a list of lengths of chromosomes in increasing order of chromosome names and separated by commas, if “Is Multiple-Chromosomes Structure” is checked. Please note that these lengths should not contain omitted regions (e.g. centromeres) in the input of chromosomes.
Run To start the reconstruction process. The main screen displays how new models are being formed from initially random models. The information about the reconstruction is displayed in the top-right corner of the main screen. The conversion factor is being used to build the model and the current value of the objective function (higher is better). After the reconstruction is finished, the score of the model is displayed in the top-right corner of the main screen (the lower the value is, the better the model is). NA
Stop During the reconstruction, if this button is pressed, the program will stop and output the currently best structure. If the program is searching for the best conversion factor, it will stop the searching and use the best-found conversion factor to build models. NA

3D model reconstruction by 3DMax

A. Purpose
To build 3D chromosomes and genome models

B. Input
A contact matrix in sparse matrix format

C. Output
3D models in .gss format file and .pdb format file

D Running
Access the function from the menu toolbar: 3D-Functions/3DMax-3D Modeler

E. 3DMax GUI

Field Description Default
Conversion Factor α in the formula d(i,j)=1/(IF(ij)^α), where IF(i,j)is interaction frequency between i and j. When the field is left blank, the program will search for the best value in the range [0.1-2.0] with a step size of 0.1. Users can also specify a range to search for by putting two numbers separated by a hyphen (e.g. 0.5-1.0). During the search, the right-top corner of the main screen displays information about the current value being tested. 0.5
Initial Learning Rate Initial learning rate of the optimization. Higher learning rate can speed up the reconstruction process but can cause the process to fail as well 1.0
Max Number of Iterations Maximum number of iterations for the optimization 2000
Chromosome Chromosome name of the contact matrix in the input. If the input contains contact matrix of the whole genome, leave this field blank. X
Genome ID Genome version of the contact matrix in the input. hg19
Is Multiple-Chromosomes Structure? if the input contains both inter-and intra-chromosomal contacts data, this checkbox should be checked. unchecked
Length of Chromosomes This field contains a list of lengths of chromosomes in increasing order of chromosome names and separated by commas, if “Is Multiple-Chromosomes Structure” is checked. Please note that these lengths should not contain omitted regions (e.g. centromeres) in the input of chromosomes.
Run To start the reconstruction process. The main screen displays how new models are being formed from initially random models. The information about the reconstruction is displayed in the top-right corner of the main screen. The conversion factor being used to build model and the current value of the objective function (higher is better). After the reconstruction is finished, the score of the model is displayed in the top-right corner of the main screen (the lower the value is, the better the model is). NA
Stop During the reconstruction, if this button is pressed, the program will stop and output the currently best structure. If the program is searching for the best conversion factor, it will stop the searching and use the best-found conversion factor to build models. NA

Chromatin loop identification

A. Purpose
To identify chromatin loop in 3D models

B. Input
A 3D model in .gss format.

C. Output
A list of chromatin loops in a .bed format file (optional) and highlighted in the 3D model

D. Running
Access the function from the menu toolbar: 3D-Functions/Loop Detection

E. Loop identification GUI

Model annotation

A. Purpose
To annotate 3D models with genomic elements

B. Input
A 3D model (e.g. in executable/sample_data/models) and genomic elements in .bed format files (e.g. in executable/track_files)

C. Output
3D model is annotated with data from .bed format files

D. Running
Access the function from the menu toolbar: 3D-Functions/Model Annotation

E. Model annotation GUI

Field Description Default
Track file A file in bed format (see executable/track_files for example) to annotate the model NA
Track name A unique name associated with the above input file Name of track file
Is domain or loop? Indicate if the track file contains domains or loops. Adjacent domains/loops will be colored in red/blue alternatively. Unchecked
Choose color To pick a color to label annotation and points overlapped by genomic elements in the track file. Random
Change color To change color of the corresponding track NA
Checking corresponding track names will display or hidden the content of tracks.

Gene expression data visualization (a special case of model annotation)

A. Purpose
To display gene expression level along a 3D model

B. Input

  • A 3D model in GSS formatt to visualize. An sample file can be found here executable/sample_data/models/chr11_10kb_gm12878_list_60mb_70mb_1514493462531.gss)
  • A gene expression data file in GCT format ( http://software.broadinstitute.org/cancer/software/genepattern/file-formats-guide#GCT ), an example file is executable/sample_data/gene_expression/allaml.dataset.gct .
  • And a text file to specify genomic coordinates of probes/genes in the GCT format file (each line consists of 4 elements separated by space or tab, e.g.: probe_or_gene_name chr_number start end). A sample is executable/sample_data/gene_expression/probe_coordinates.txt .

    These following 3 files are prepared for demo:
  • executable/sample_data/models/ chr11_10kb_gm12878_list_60mb_70mb_1514493462531.gss
  • executable/sample_data/gene_expression/allaml.dataset.gct
  • executable/sample_data/gene_expression/ probe_coordinates.txt.

C. Output
Expression levels of genes/probes are annotated in the 3D model. Usually, the GCT file contains several samples and therefore, the median value (across all samples) together with minimum and maximum values (in brackets) are displayed next to probe/gene names. If the 3D model and the gene expression data file have no overlap, no annotation will be added to the 3D model.

D. Running
Access the function from the menu toolbar: 3D-Functions/Model Annotation. A GCT file must be filled in the “Track File” field.

E. Gene Expression Model annotation GUI

Comparing 2 models

A. Purpose
To superimpose and compare two 3D-models in GSS format.

B. Input
Two chromosome models in GSS format.

C. Output
The two models are scaled, superimposed and visualized. Spearman’s correlation and RMSE between pairwise distances of the two models are calculated.

D. Running
Access the function from the menu toolbar: 3D-Functions/Compare Models

E. Model Comparison GUI