1.0.0

Version 1.0.0

Major Underlying Changes

• Integration of Ibex with immApex
• Updated Seurat object to v5
• Runs using basilisk instead of reticulate - no installation of python packages

Feature Changes

• Converted Ibex.matrix() to Ibex_matrix()
• Updated support for SCE format for runIbex()
• Update CoNGAfy() to function with all versions of Seurat
• Updated quietBCRgenes() to use VariableFeatures() call for SeuratV5 and backward compatibility.
• Add getHumanIgPseudoGenes() to return a list of human Immunoglobulin Pseudo genes that are kept by quietBCRgenes()

New Models

• Added New Light and Heavy Chain Models
• Encoding methods now accepted: "OHE", "atchleyFactors", "crucianiProperties", "kideraFactors", "MSWHIM","tScales", "zScales"
• Sequence input:
  • Human Heavy: 10000000
  • Human Light: 5000000
  • Human Heavy-Expanded: 5000000
  • Human Light-Expanded: 2500000
  • Mouse Heavy: 5000000
  • Mouse Heavy-Expanded: 5000000
• Trained convolutional and variational autoencoders for Heavy/Light chains
  • Architecture: 512-256-128-256-512
  • Parameters:
    Batch Size = 128
    Latent Dimensions = 128
    Epochs = 100
    Loss = Mean Squared Error (CNN) & KL Divergence (VAE)
    Activation = relu
    Learning rate = 1e-6
  • Optimizers: Adam
  • Early stopping was set to patients of 10 for minimal validation loss and restoration of best weights
  • CNN autoencoders have batch normalization layers between the dense layers.