Brobustedger

Brobustedger-package is developed to identify differentially expressed genes (DEGs) from RNA-seq count data through boosting robust edgeR when the dataset contains outlier/noise or missing values. Iterative leave-out (iLOO) method is used to detect of gene specific outliers and random forest is used to impute null/missing values. Finally DEGs are calculated by edgeR (Robust). Brobustedger-package provides a user-friendly R-package interface. Users can utilize its DEGList function to find differentially expressed gene list. Additionally, Brobustedger-package offers functionalities for outlier detection (iLOO) and imputation of missing or null values (NI) by using iLOO and random forest method, respectively.

Installation

#Install edgeR from bioconductor
if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("edgeR")
# Install devtools from CRAN
install.packages("devtools")

# development version from GitHub:
require("devtools")
devtools::install_github("BandhanSarker/Brobustedger")

Data Pre-process

Step-1:

iLOO is used to identify gene specific outliers from RNA-seq count data.

Usage

iLOO(data)

Arguments

• data RNA-seq count data matrix, whose row contain genes and column contain samples

Value

Output of the function identify gene specific outliers. Non-outlier values are displayed as "NA"

Examples

Brobustedger::iLOO(sampleDataOut) # Input RNA-seq count data

Step-2:

NI is used for the imputation of null/missing values in RNA-seq count data.

Usage

NI(data, mi = 5, NT = 100)

Arguments

• data RNA-seq count data matrix contains missing/Null value, whose row contain genes and column contain samples
• mi maximum number of iterations to impute nullify values
• NT number of trees in each set to impute nullify values

Value

Clean dataset after the imputation of missing/nullify value

Examples

Brobustedger::NI(sampleDataMiss)  #Input RNA-seq count data with missing or null value

Step-3:

DEGList is used to identify differentially expressed genes from RNA-seq count data.

Usage

#Identify differentially expressed gene list
DEGList (data, n1, n2, p.threshold = 0.05)

Arguments

• data RNA-seq count data matrix, whose row contains genes and column contains samples
• n1 control group
• n2 case group
• p.threshold adjusted p-value by default=0.05
• padjust select the p-value adjustment method by default="BH". choose the possible method "none", "BH", "fdr", "BY" and "holm". See help section of BrobustedgeR-package for more details.
• mi maximum number of iterations to impute nullify values by default= 5
• NT number of trees in each set to impute nullify values by default=10

Value

1. Gene.list: Differentially expressed genes list and 2. result: others output

Examples

 data (sampleDataOut)
 DEGList(sampleDataOut,3,3,0.05)

Example Dataset

#Sample RNA-seq count data with outliers value
data(sampleDataOut)

#Sample RNA-seq count data with missing value
data(sampleDataMiss)

Author(s):

Bandhan Sarker, Md. Matiur Rahaman, Muhammad Habibulla Alamin, Md. Ariful Islam and Md. Nurul Haque Mollah

References

Robinson MD, McCarthy DJ and Smyth GK (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139-140

Stekhoven, D.J. and Buehlmann, P. (2012), 'MissForest - nonparametric missing value imputation for mixed-type data', Bioinformatics, 28(1) 2012, 112-118, doi: 10.1093/bioinformatics/btr597

George NI, Bowyer JF, Crabtree NM, Chang C-W (2015) An Iterative Leave-One-Out Approach to Outlier Detection in RNA-Seq Data. PLoS ONE 10(6): e0125224. https://doi.org/10.1371/journal.pone.0125224