explore_batch_effect: Explore batch effect.

In markgene/yamatClassifier: Yet Another Methylation Array Toolkit: Classifier

Explore batch effect.

Description

Investigate if batch effect exists.

Usage

explore_batch_effect(
  trainer,
  batch_name = "Batch",
  classification_name = "Classification",
  top_n_rle = 20000,
  rle_downsample = 100,
  top_n_pca = 20000,
  k = 50,
  threshold = 0.9
)

Arguments

trainer	A S3 object of YamatClassifierTrainer class.
batch_name	column name of batch in the targets attribute of YamatClassifierTrainer object.
classification_name	column name of classification in the targets attribute of YamatClassifierTrainer object.
top_n_rle	an integer of the most variable N loci for relative log expression (RLE) analysis.
rle_downsample	if there are many samples, down-sample it to a number while keep the proportion of each group.
top_n_pca	an integer of the most variable N loci for PCA.
k	Number of eigenvalues requested.
threshold	A numeric scalar between 0 to 1 of the threshold of the fraction of variance to choose PC number. Default to 0.9.

Details

1. Center and scale.

2. Compute the correlation/covariance matrix.

3. Calculate the eigenvectors and eigenvalues.

4. Choose the PC number. I use Capper's method and fraction of variance to calculate PC numbers and choose the bigger one from the two methods. See details at find_pc_number.capper and find_pc_number.var_frac.

5. Project the scaled input matrix onto the new basis.

markgene/yamatClassifier documentation built on Oct. 14, 2024, 2:36 a.m.