maTrex: Main Trex interface

In ncborcherding/Trex: Methods for TCR single-cell embedding

Main Trex interface

Description

Use this to run the Trex algorithm to return latent vectors in the form of a matrix or if you prefer.

Usage

maTrex(
  sc,
  chains = "TRA",
  method = "encoder",
  encoder.model = "VAE",
  encoder.input = "AF",
  theta = pi
)

Arguments

sc	Single-cell object or the output of combineTCR() in scRepertoire
chains	TCR chain to use - TRA or TRB.
method	"encoder" = using deep learning autoencoders or "geometric" = geomteric transformations based on BLOSUM62 matrix
encoder.model	"AE" = dense autoencoder or "VAE" = variation autoencoder
encoder.input	Atchley factors (AF), Kidera factors ((AF)), or or One-Hot Encoder (OHE).
theta	angle to use for geometric transformation

Value

Trex encoded values from the autoencoder

Examples

Trex_values <- maTrex(trex_example, 
                     chains = "TRA",
                     method = "encoder",
                     encoder.model = "VAE",
                     encoder.input = "AF")
                          
Trex_values <- maTrex(trex_example, 
                     chains = "TRA",
                     method = "geometric",
                     theta = pi)
                        

ncborcherding/Trex documentation built on Nov. 4, 2024, 10:31 p.m.