BERT (Batch-Effect Removal with Trees) offers flexible and efficient batch effect correction of omics data, while providing maximum tolerance to missing values. Tested on multiple datasets from proteomic analyses, BERT offered a typical 5-10x runtime improvement over existing methods, while retaining more numeric values and preserving batch effect reduction quality.
As such, BERT is a valuable preprocessing tool for data analysis workflows, in particular for proteomic data. By providing BERT via Bioconductor, we make this tool available to a wider research community. An accompanying research paper is currently under preparation and will be made public soon.
BERT addresses the same fundamental data integration challenges than the [HarmonizR][https://github.com/HSU-HPC/HarmonizR] package, which is released on Bioconductor in November 2023. However, various algorithmic modications and optimizations of BERT provide better execution time and better data coverage than HarmonizR. Moreover, BERT offers a more user-friendly design and a less error-prone input format.
Please note that our package BERT is neither affiliated with nor related to Bidirectional Encoder Representations from Transformers as published by Google.
Please report any questions and issues in the GitHub forum, the BioConductor forum or directly contact the authors,
Please download and install a current version of R (Windows binaries). You might want to consider installing a development environment as well, e.g. RStudio. Finally, BERT can be installed via Bioconductor using
if (!require("BiocManager", quietly = TRUE)){
install.packages("BiocManager")
}
BiocManager::install("BERT")
which will install all required dependencies. To install the development version of BERT, you can use devtools as follows
which may require the manual installation of the dependencies
sva
and limma
.
As input, BERT requires a dataframe1 with samples in rows
and features in columns. For each sample, the respective batch should be
indicated by an integer or string in a corresponding column labelled
Batch. Missing values should be labelled as NA
. A
valid example dataframe could look like this:
example = data.frame(feature_1 = stats::rnorm(5), feature_2 = stats::rnorm(5), Batch=c(1,1,2,2,2))
example
#> feature_1 feature_2 Batch
#> 1 0.46280422 0.2435380 1
#> 2 1.40177881 -0.1009406 1
#> 3 -0.92837904 -0.7772351 2
#> 4 -0.04259076 -0.2930548 2
#> 5 0.35043118 -0.6720355 2
Note that each batch should contain at least two samples. Optional columns that can be passed are
Label
A column with integers or strings indicating
the (known) class for each sample. NA
is not allowed. BERT
may use this columns and Batch
to compute quality metrics
after batch effect correction.
Sample
A sample name. This column is ignored by BERT
and can be used to provide meta-information for further
processing.
Cov_1
, Cov_2
, …, Cov_x
:
One or multiple columns with integers, indicating one or several
covariate levels. NA
is not allowed. If this(these)
column(s) is present, BERT will pass them as covariates to the the
underlying batch effect correction method. As an example, this
functionality can be used to preserve differences between
healthy/tumorous samples, if some of the batches exhibit strongly
variable class distributions. Note that BERT requires at least two
numeric values per batch and unique covariate level to adjust a feature.
Features that don’t satisfy this condition in a specific batch are set
to NA
for that batch.
Reference
A column with integers or strings from
ℕ0 that indicate, whether a
sample should be used for “learning” the transformation for batch effect
correction or whether the sample should be co-adjusted using the learned
transformation from the other samples.NA
is not allowed.
This feature can be used, if some batches contain unique classes or
samples with unknown classes which would prohibit the usage of covariate
columns. If the column contains a 0
for a sample, this
sample will be co-adjusted. Otherwise, the sample should contain the
respective class (encoded as integer or string). Note that BERT requires
at least two references of common class per adjustment step and that the
Reference
column is mutually exclusive with covariate
columns.
Note that BERT tries to find all metadata information for a
SummarizedExperiment
, including the mandatory batch
information, using colData
. For instance, a valid
SummarizedExperiment
might be defined as
nrows <- 200
ncols <- 8
expr_values <- matrix(runif(nrows * ncols, 1, 1e4), nrows)
# colData also takes all other metadata information, such as Label, Sample,
# Covariables etc.
colData <- data.frame(Batch=c(1,1,1,1,2,2,2,2), Reference=c(1,1,0,0,1,1,0,0))
dataset_raw = SummarizedExperiment::SummarizedExperiment(assays=list(expr=expr_values), colData=colData)
BERT can be invoked by importing the BERT
library and
calling the BERT
function. The batch effect corrected data
is returned as a dataframe that mirrors the input dataframe2.
library(BERT)
# generate test data with 10% missing values as provided by the BERT library
dataset_raw <- generate_dataset(features=60, batches=10, samplesperbatch=10, mvstmt=0.1, classes=2)
# apply BERT
dataset_adjusted <- BERT(dataset_raw)
#> 2024-09-21 02:53:33.719112 INFO::Formatting Data.
#> 2024-09-21 02:53:33.727101 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:33.735636 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:34.335616 INFO::Found 600 missing values.
#> 2024-09-21 02:53:34.346112 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:34.346665 INFO::Done
#> 2024-09-21 02:53:34.347093 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:34.357635 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:34.358368 INFO::Found 10 batches.
#> 2024-09-21 02:53:34.358759 INFO::Cores argument is not defined or BPPARAM has been specified. Argument corereduction will not be used.
#> 2024-09-21 02:53:36.222617 INFO::Using default BPPARAM
#> 2024-09-21 02:53:36.223129 INFO::Processing subtree level 1
#> 2024-09-21 02:53:37.593243 INFO::Processing subtree level 2
#> 2024-09-21 02:53:38.939775 INFO::Adjusting the last 1 batches sequentially
#> 2024-09-21 02:53:38.941575 INFO::Done
#> 2024-09-21 02:53:38.942168 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:38.946945 INFO::ASW Batch was 0.527558781388249 prior to batch effect correction and is now -0.137914832133978 .
#> 2024-09-21 02:53:38.947609 INFO::ASW Label was 0.295186424280582 prior to batch effect correction and is now 0.783382207202106 .
#> 2024-09-21 02:53:38.948588 INFO::Total function execution time is 5.24417185783386 s and adjustment time is 4.58343458175659 s ( 87.4 )
BERT uses the logging
library to convey live information
to the user during the adjustment procedure. The algorithm first
verifies the shape and suitability of the input dataframe (lines 1-6)
before continuing with the actual batch effect correction (lines 8-14).
BERT measure batch effects before and after the correction step by means
of the average silhouette score (ASW) with respect to batch and labels
(lines 7 and 15). The ASW Label should increase in a successful batch
effect correction, whereas low values ( ≤ 0) are desireable for the ASW Batch3. Finally,
BERT prints the total function execution time (including the computation
time for the quality metrics).
BERT offers a large number of parameters to customize the batch effect adjustment. The full function call, including all defaults is
BERT(data, cores = NULL, combatmode = 1, corereduction=2, stopParBatches=2, backend="default", method="ComBat", qualitycontrol=TRUE, verify=TRUE, labelname="Label", batchname="Batch", referencename="Reference", samplename="Sample", covariatename=NULL, BPPARAM=NULL, assayname=NULL)
In the following, we list the respective meaning of each parameter: -
data
: The input dataframe/matrix/SummarizedExperiment to
adjust. See Data Preparation for
detailed formatting instructions. - data
The data for
batch-effect correction. Must contain at least two samples per batch and
2 features.
cores
: BERT uses BiocParallel
for parallelization. If the user specifies a value cores
,
BERT internally creates and uses a new instance of
BiocParallelParam
, which is however not exhibited to the
user. Setting this parameter can speed up the batch effect adjustment
considerably, in particular for large datasets and on unix-based
operating systems. A value between 2
and 4 is a reasonable choice for
typical commodity hardware. Multi-node computations are not supported as
of now. If, however, cores
is not specified, BERT will
default to BiocParallel::bpparam()
, which may have been set
by the user or the system. Additionally, the user can directly specify a
specific instance of BiocParallelParam
to be used via the
BPPARAM
argument.combatmode
An integer that encodes the parameters to
use for ComBat.Value | par.prior | mean.only |
---|---|---|
1 | TRUE | FALSE |
2 | TRUE | TRUE |
3 | FALSE | FALSE |
4 | FALSE | TRUE |
The value of this parameter will be ignored, if
method!="ComBat"
.
corereduction
Positive integer indicating the factor
by which the number of processes should be reduced, once no further
adjustment is possible for the current number of batches.4 This parameter is used
only, if the user specified a custom value for parameter
cores
.
stopParBatches
Positive integer indicating the
minimum number of batches required at a hierarchy level to proceed with
parallelized adjustment. If the number of batches is smaller, adjustment
will be performed sequentially to avoid communication
overheads.
backend
: The backend to use for inter-process
communication. Possible choices are default
and
file
, where the former refers to the default communication
backend of the requested parallelization mode and the latter will create
temporary .rds
files for data communication. ‘default’ is
usually faster for small to medium sized datasets.
method
: The method to use for the underlying batch
effect correction steps. Should be either ComBat
,
limma
for limma::removeBatchEffects
or
ref
for adjustment using specified references (cf. Data Preparation). The underlying batch
effect adjustment method for ref
is a modified version of
the limma
method.
qualitycontrol
: A boolean to (de)activate the ASW
computation. Deactivating the ASW computations accelerates the
computations.
verify
: A boolean to (de)activate the initial format
check of the input data. Deactivating this verification step accelerates
the computations.
labelname
: A string containing the name of the
column to use as class labels. The default is “Label”.
batchname
: A string containing the name of the
column to use as batch labels. The default is “Batch”.
referencename
: A string containing the name of the
column to use as reference labels. The default is “Reference”.
covariatename
: A vector containing the names of
columns with categorical covariables.The default is NULL, in which case
all column names are matched agains the pattern “Cov”.
BPPARAM
: An instance of
BiocParallelParam
that will be used for parallelization.
The default is null, in which case the value of cores
determines the behaviour of BERT.
assayname
: If the user chooses to pass a
SummarizedExperiment
object, they need to specify the name
of the assay that they want to apply BERT to here. BERT then returns the
input SummarizedExperiment
with an additional assay labeled
assayname_BERTcorrected
.
BERT utilizes the logging
package for output. The user
can easily specify the verbosity of BERT by setting the global logging
level in the script. For instance
BERT exhibits a large number of parameters for parallelisation as to provide users with maximum flexibility. For typical scenarios, however, the default parameters are well suited. For very large experiments ( > 15 batches), we recommend to increase the number of cores (a reasonable value is 4 but larger values may be possible on your hardware). Most users should leave all parameters to their respective default.
In the following, we present simple cookbook examples for BERT usage. Note that ASWs (and runtime) will most likely differ on your machine, since the data generating process involves multiple random choices.
Here, BERT uses limma as underlying batch effect correction algorithm
(method='limma'
) and performs all computations on a single
process (cores
parameter is left on default).
# import BERT
library(BERT)
# generate data with 30 batches, 60 features, 15 samples per batch, 15% missing values and 2 classes
dataset_raw <- generate_dataset(features=60, batches=20, samplesperbatch=15, mvstmt=0.15, classes=2)
# BERT
dataset_adjusted <- BERT(dataset_raw, method="limma")
#> 2024-09-21 02:53:38.996813 INFO::Formatting Data.
#> 2024-09-21 02:53:38.997432 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:38.998234 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:38.999976 INFO::Found 2700 missing values.
#> 2024-09-21 02:53:39.018309 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:39.018866 INFO::Done
#> 2024-09-21 02:53:39.019238 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:39.027565 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:39.028143 INFO::Found 20 batches.
#> 2024-09-21 02:53:39.028533 INFO::Cores argument is not defined or BPPARAM has been specified. Argument corereduction will not be used.
#> 2024-09-21 02:53:39.028948 INFO::Using default BPPARAM
#> 2024-09-21 02:53:39.02931 INFO::Processing subtree level 1
#> 2024-09-21 02:53:39.356365 INFO::Processing subtree level 2
#> 2024-09-21 02:53:39.665008 INFO::Processing subtree level 3
#> 2024-09-21 02:53:39.982679 INFO::Adjusting the last 1 batches sequentially
#> 2024-09-21 02:53:39.984071 INFO::Done
#> 2024-09-21 02:53:39.984549 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:39.992996 INFO::ASW Batch was 0.458005658708713 prior to batch effect correction and is now -0.123419352639953 .
#> 2024-09-21 02:53:39.993566 INFO::ASW Label was 0.317002444524643 prior to batch effect correction and is now 0.826539061435857 .
#> 2024-09-21 02:53:39.994247 INFO::Total function execution time is 0.997483968734741 s and adjustment time is 0.956056356430054 s ( 95.85 )
Here, BERT uses ComBat as underlying batch effect correction
algorithm (method
is left on default) and performs all
computations on a 2 processes (cores=2
).
# import BERT
library(BERT)
# generate data with 30 batches, 60 features, 15 samples per batch, 15% missing values and 2 classes
dataset_raw <- generate_dataset(features=60, batches=20, samplesperbatch=15, mvstmt=0.15, classes=2)
# BERT
dataset_adjusted <- BERT(dataset_raw, cores=2)
#> 2024-09-21 02:53:40.023169 INFO::Formatting Data.
#> 2024-09-21 02:53:40.023826 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:40.024614 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:40.026433 INFO::Found 2700 missing values.
#> 2024-09-21 02:53:40.045974 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:40.046583 INFO::Done
#> 2024-09-21 02:53:40.046947 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:40.055599 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:40.056242 INFO::Found 20 batches.
#> 2024-09-21 02:53:40.583478 INFO::Set up parallel execution backend with 2 workers
#> 2024-09-21 02:53:40.584523 INFO::Processing subtree level 1 with 20 batches using 2 cores.
#> 2024-09-21 02:53:42.662623 INFO::Adjusting the last 2 batches sequentially
#> 2024-09-21 02:53:42.663687 INFO::Adjusting sequential tree level 1 with 2 batches
#> 2024-09-21 02:53:43.779245 INFO::Done
#> 2024-09-21 02:53:43.779774 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:43.787508 INFO::ASW Batch was 0.481333932909203 prior to batch effect correction and is now -0.131521031221386 .
#> 2024-09-21 02:53:43.787936 INFO::ASW Label was 0.295040470919317 prior to batch effect correction and is now 0.853881383224244 .
#> 2024-09-21 02:53:43.788487 INFO::Total function execution time is 3.76541757583618 s and adjustment time is 3.72284126281738 s ( 98.87 )
Here, BERT takes the input data using a
SummarizedExperiment
instead. Batch effect correction is
then performed using ComBat as underlying algorithm (method
is left on default) and all computations are performed on a single
process (cores
parameter is left on default).
nrows <- 200
ncols <- 8
# SummarizedExperiments store samples in columns and features in rows (in contrast to BERT).
# BERT will automatically account for this.
expr_values <- matrix(runif(nrows * ncols, 1, 1e4), nrows)
# colData also takes further metadata information, such as Label, Sample,
# Reference or Covariables
colData <- data.frame("Batch"=c(1,1,1,1,2,2,2,2), "Label"=c(1,2,1,2,1,2,1,2), "Sample"=c(1,2,3,4,5,6,7,8))
dataset_raw = SummarizedExperiment::SummarizedExperiment(assays=list(expr=expr_values), colData=colData)
dataset_adjusted = BERT(dataset_raw, assayname = "expr")
#> 2024-09-21 02:53:43.830459 INFO::Formatting Data.
#> 2024-09-21 02:53:43.830977 INFO::Recognized SummarizedExperiment
#> 2024-09-21 02:53:43.83131 INFO::Typecasting input to dataframe.
#> 2024-09-21 02:53:43.853751 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:43.854574 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:43.856658 INFO::Found 0 missing values.
#> 2024-09-21 02:53:43.860569 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:43.860922 INFO::Done
#> 2024-09-21 02:53:43.861249 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:43.863757 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:43.864211 INFO::Found 2 batches.
#> 2024-09-21 02:53:43.864531 INFO::Cores argument is not defined or BPPARAM has been specified. Argument corereduction will not be used.
#> 2024-09-21 02:53:43.864883 INFO::Using default BPPARAM
#> 2024-09-21 02:53:43.865183 INFO::Adjusting the last 2 batches sequentially
#> 2024-09-21 02:53:43.865764 INFO::Adjusting sequential tree level 1 with 2 batches
#> 2024-09-21 02:53:43.900294 INFO::Done
#> 2024-09-21 02:53:43.900845 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:43.903992 INFO::ASW Batch was -0.011859070358952 prior to batch effect correction and is now -0.100839603753655 .
#> 2024-09-21 02:53:43.904435 INFO::ASW Label was -0.00585204030368612 prior to batch effect correction and is now 0.0074723333953661 .
#> 2024-09-21 02:53:43.904996 INFO::Total function execution time is 0.0745351314544678 s and adjustment time is 0.0361301898956299 s ( 48.47 )
BERT can utilize categorical covariables that are specified in
columns Cov_1, Cov_2, ...
. These columns are automatically
detected and integrated into the batch effect correction process.
# import BERT
library(BERT)
# set seed for reproducibility
set.seed(1)
# generate data with 5 batches, 60 features, 30 samples per batch, 15% missing values and 2 classes
dataset_raw <- generate_dataset(features=60, batches=5, samplesperbatch=30, mvstmt=0.15, classes=2)
# create covariable column with 2 possible values, e.g. male/female condition
dataset_raw["Cov_1"] = sample(c(1,2), size=dim(dataset_raw)[1], replace=TRUE)
# BERT
dataset_adjusted <- BERT(dataset_raw)
#> 2024-09-21 02:53:43.928598 INFO::Formatting Data.
#> 2024-09-21 02:53:43.929159 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:43.929902 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:43.931499 INFO::Found 1350 missing values.
#> 2024-09-21 02:53:43.932151 INFO::BERT requires at least 2 numeric values per batch/covariate level. This may reduce the number of adjustable features considerably, depending on the quantification technique.
#> 2024-09-21 02:53:43.94689 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:43.947431 INFO::Done
#> 2024-09-21 02:53:43.947817 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:43.951638 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:43.952169 INFO::Found 5 batches.
#> 2024-09-21 02:53:43.952579 INFO::Cores argument is not defined or BPPARAM has been specified. Argument corereduction will not be used.
#> 2024-09-21 02:53:43.953011 INFO::Using default BPPARAM
#> 2024-09-21 02:53:43.953401 INFO::Processing subtree level 1
#> 2024-09-21 02:53:44.134101 INFO::Adjusting the last 2 batches sequentially
#> 2024-09-21 02:53:44.135688 INFO::Adjusting sequential tree level 1 with 2 batches
#> 2024-09-21 02:53:44.206882 INFO::Done
#> 2024-09-21 02:53:44.207449 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:44.211257 INFO::ASW Batch was 0.492773245691086 prior to batch effect correction and is now -0.0377157224767566 .
#> 2024-09-21 02:53:44.211675 INFO::ASW Label was 0.40854766060101 prior to batch effect correction and is now 0.895560693013661 .
#> 2024-09-21 02:53:44.212279 INFO::Total function execution time is 0.283723115921021 s and adjustment time is 0.254772901535034 s ( 89.8 )
In rare cases, class distributions across experiments may be severely
skewed. In particular, a batch might contain classes that other batches
don’t contain. In these cases, samples of common conditions may serve as
references (bridges) between the batches
(method="ref"
). BERT utilizes those samples as references
that have a condition specified in the “Reference” column of the input.
All other samples are co-adjusted. Please note, that this strategy
implicitly uses limma as underlying batch effect correction
algorithm.
# import BERT
library(BERT)
# generate data with 4 batches, 6 features, 15 samples per batch, 15% missing values and 2 classes
dataset_raw <- generate_dataset(features=6, batches=4, samplesperbatch=15, mvstmt=0.15, classes=2)
# create reference column with default value 0. The 0 indicates, that the respective sample should be co-adjusted only.
dataset_raw[, "Reference"] <- 0
# randomly select 2 references per batch and class - in practice, this choice will be determined by external requirements (e.g. class known for only these samples)
batches <- unique(dataset_raw$Batch) # all the batches
for(b in batches){ # iterate over all batches
# references from class 1
ref_idx = sample(which((dataset_raw$Batch==b)&(dataset_raw$Label==1)), size=2, replace=FALSE)
dataset_raw[ref_idx, "Reference"] <- 1
# references from class 2
ref_idx = sample(which((dataset_raw$Batch==b)&(dataset_raw$Label==2)), size=2, replace=FALSE)
dataset_raw[ref_idx, "Reference"] <- 2
}
# BERT
dataset_adjusted <- BERT(dataset_raw, method="ref")
#> 2024-09-21 02:53:44.250179 INFO::Formatting Data.
#> 2024-09-21 02:53:44.250799 INFO::Replacing NaNs with NAs.
#> 2024-09-21 02:53:44.251446 INFO::Removing potential empty rows and columns
#> 2024-09-21 02:53:44.252149 INFO::Found 60 missing values.
#> 2024-09-21 02:53:44.254772 INFO::Introduced 0 missing values due to singular proteins at batch/covariate level.
#> 2024-09-21 02:53:44.255135 INFO::Done
#> 2024-09-21 02:53:44.255513 INFO::Acquiring quality metrics before batch effect correction.
#> 2024-09-21 02:53:44.257568 INFO::Starting hierarchical adjustment
#> 2024-09-21 02:53:44.258051 INFO::Found 4 batches.
#> 2024-09-21 02:53:44.258421 INFO::Cores argument is not defined or BPPARAM has been specified. Argument corereduction will not be used.
#> 2024-09-21 02:53:44.258821 INFO::Using default BPPARAM
#> 2024-09-21 02:53:44.259162 INFO::Processing subtree level 1
#> 2024-09-21 02:53:44.337838 INFO::Adjusting the last 2 batches sequentially
#> 2024-09-21 02:53:44.339406 INFO::Adjusting sequential tree level 1 with 2 batches
#> 2024-09-21 02:53:44.361279 INFO::Done
#> 2024-09-21 02:53:44.361838 INFO::Acquiring quality metrics after batch effect correction.
#> 2024-09-21 02:53:44.364414 INFO::ASW Batch was 0.440355021914032 prior to batch effect correction and is now -0.087480278736629 .
#> 2024-09-21 02:53:44.364899 INFO::ASW Label was 0.373906827748893 prior to batch effect correction and is now 0.919791677398366 .
#> 2024-09-21 02:53:44.36565 INFO::Total function execution time is 0.115540027618408 s and adjustment time is 0.10325813293457 s ( 89.37 )
Issues can be reported in the GitHub forum, the BioConductor forum or directly to the authors.
This code is published under the GPLv3.0 License and is available for non-commercial academic purposes.
Please cite our manuscript, if you use BERT for your research: Yannis Schumann, Simon Schlumbohm et al., BERT - Batch Effect Reduction Trees with Missing Value Tolerance, 2023
sessionInfo()
#> R version 4.4.1 (2024-06-14)
#> Platform: x86_64-pc-linux-gnu
#> Running under: Ubuntu 24.04.1 LTS
#>
#> Matrix products: default
#> BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
#> LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.26.so; LAPACK version 3.12.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=C
#> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
#> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
#> [9] LC_ADDRESS=C LC_TELEPHONE=C
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#>
#> time zone: Etc/UTC
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> other attached packages:
#> [1] BERT_1.1.2 BiocStyle_2.33.1
#>
#> loaded via a namespace (and not attached):
#> [1] tidyselect_1.2.1 blob_1.2.4
#> [3] Biostrings_2.73.1 fastmap_1.2.0
#> [5] janitor_2.2.0 XML_3.99-0.17
#> [7] digest_0.6.37 timechange_0.3.0
#> [9] lifecycle_1.0.4 cluster_2.1.6
#> [11] statmod_1.5.0 survival_3.7-0
#> [13] KEGGREST_1.45.1 invgamma_1.1
#> [15] RSQLite_2.3.7 magrittr_2.0.3
#> [17] genefilter_1.87.0 compiler_4.4.1
#> [19] rlang_1.1.4 sass_0.4.9
#> [21] tools_4.4.1 yaml_2.3.10
#> [23] knitr_1.48 S4Arrays_1.5.7
#> [25] bit_4.5.0 DelayedArray_0.31.11
#> [27] abind_1.4-8 BiocParallel_1.39.0
#> [29] BiocGenerics_0.51.1 sys_3.4.2
#> [31] grid_4.4.1 stats4_4.4.1
#> [33] xtable_1.8-4 edgeR_4.3.15
#> [35] iterators_1.0.14 logging_0.10-108
#> [37] SummarizedExperiment_1.35.1 cli_3.6.3
#> [39] rmarkdown_2.28 crayon_1.5.3
#> [41] generics_0.1.3 httr_1.4.7
#> [43] DBI_1.2.3 cachem_1.1.0
#> [45] stringr_1.5.1 zlibbioc_1.51.1
#> [47] splines_4.4.1 parallel_4.4.1
#> [49] AnnotationDbi_1.67.0 BiocManager_1.30.25
#> [51] XVector_0.45.0 matrixStats_1.4.1
#> [53] vctrs_0.6.5 Matrix_1.7-0
#> [55] jsonlite_1.8.9 sva_3.53.0
#> [57] comprehenr_0.6.10 IRanges_2.39.2
#> [59] S4Vectors_0.43.2 bit64_4.0.5
#> [61] maketools_1.3.0 locfit_1.5-9.10
#> [63] foreach_1.5.2 limma_3.61.9
#> [65] jquerylib_0.1.4 annotate_1.83.0
#> [67] glue_1.7.0 codetools_0.2-20
#> [69] lubridate_1.9.3 stringi_1.8.4
#> [71] GenomeInfoDb_1.41.1 GenomicRanges_1.57.1
#> [73] UCSC.utils_1.1.0 htmltools_0.5.8.1
#> [75] GenomeInfoDbData_1.2.12 R6_2.5.1
#> [77] evaluate_1.0.0 lattice_0.22-6
#> [79] Biobase_2.65.1 png_0.1-8
#> [81] memoise_2.0.1 snakecase_0.11.1
#> [83] bslib_0.8.0 SparseArray_1.5.36
#> [85] nlme_3.1-166 mgcv_1.9-1
#> [87] xfun_0.47 MatrixGenerics_1.17.0
#> [89] buildtools_1.0.0
Matrices and SummarizedExperiments work as well, but will automatically be converted to dataframes.↩︎
In particular, the row and column names are in the same order and the optional columns are preserved.↩︎
The optimum of ASW Label is 1, which is typically however not achieved on real-world datasets. Also, the optimum of ASW Batch can vary, depending on the class distributions of the batches.↩︎
E.g. consider a BERT call with 8 batches and 8
processes. Further adjustment is not possible with this number of
processes, since batches are always processed in pairs. With
corereduction=2
, the number of processes for the following
adjustment steps would be set to 8/2 = 4, which is the maximum number of
usable processes for this example.↩︎