5.2 EMP_adjust_abudance

Batch effect refers to systematic variation caused by various factors during data collection, which is independent of the study variable. Batch effects are usually caused by the following reasons: different parts of the experiment are completed at different times, changes in experimental conditions (including instruments, reagent batches, reagent dosage, sequencing batches, etc.), different data sources (for example: integrating your own data set with public data sets for analysis), etc. The existence of batch effects may affect data analysis and result interpretation, and ordinary standardization methods are not enough to adjust the differences between batches. The module EMP_adjust_abudance can help users effectively solve the correction problem of batch effects. It contains the three most commonly used correction methods: combat_seq, combat, and limma_remove_batch_effect. These methods can improve the comparability and reliability of data by reducing batch effects.

Note：
Before performing batch calibration, it is necessary to integrate the project data from different batches into a single data file.

5.2.1 Combat method

This module references the comBat algorithm of the sva package to correct batch effects. comBat uses parametric or nonparametric empirical Bayesian models to correct batch effects. The input data is the cleaned and standardized expression data, and the returned data is an expression matrix that has been corrected for batch effects.

🏷️Example：

Use module EMP_assay_extract to extract the assay of geno_ko . Utilize parameter .factor_unwanted within module EMP_adjust_abudance to specify factors of no interest (factors requiring batch correction) as Region , parameter .factor_of_interest to specify factors of interest as 'Group' , and parameter method to specify the use of 'combat' method for batch effect correction.

MAE |>
  EMP_assay_extract(experiment='geno_ko') |>
  EMP_adjust_abundance(.factor_unwanted = 'Region',.factor_of_interest = 'Group',
                      method = 'combat')

5.2.2 Combat_seq method

This module uses the module comBat_seq of the sva package to correct batch effects. comBat_seq is an improved model of combat, using negative binomial regression, specifically for RNA-Seq count data.

🏷️Example：

MAE |>
  EMP_assay_extract(experiment='geno_ko') |>
  EMP_adjust_abundance(.factor_unwanted = 'Region',.factor_of_interest = 'Group',
                      method = 'combat_seq')

5.2.3 Limma_remove_batch_effect method

This module references the module removeBatch Effect of the limma package to correct batch effects. This module is typically used to remove batch effects from microarray data or RNA sequencing data.

🏷️Example：

MAE |>
  EMP_assay_extract(experiment='geno_ko') |>
  EMP_adjust_abundance(.factor_unwanted = 'Region',.factor_of_interest = 'Group',
                      method = 'limma_remove_batch_effect')

5.2 EMP_adjust_abudance