Function reference • TBSignatureProfiler

Signature Profiling

Functions and data used for profiling RNA-seq data or organizing profiled data.

runTBsigProfiler(): Run TB gene signature profiling.

TBSPapp(): Run the TBSignatureProfiler Shiny application.

compareAlgs(): Compare scoring algorithms on a single signature via heatmap or boxplot.

mkAssay(): Add SummarizedExperiment assays to the data structure.

TBsignatures: A list of published TB signatures.

TBcommon: A list of published TB signatures, using author-given names.

sigAnnotData: Annotation information for published TB signatures.

common_sigAnnotData: Annotation information for published TB signatures.

COVIDsignatures: A list of published/pre-print COVID-19 signatures.

addTBsignature(): Introduce a new signature into the TBSignatureProfiler.

TBsignaturesSplit: Up/Down-regulated genes information for selected TB signatures.

Plotting Profiled Signatures

Functions to plot profiled RNA-seq data and compare signatures.

signatureHeatmap(): Plot a heatmap of signature scores.

signatureGeneHeatmap(): Plot a heatmap of a single signature score with individual gene expression levels.

signatureBoxplot(): Plot a boxplot of signature genes.

distinctColors(): Generate a distinct palette for coloring different clusters.

Signature Evaluation

Functions to evaluate signatures’ predictive accuracy via AUC and ROC curve calculations.

bootstrapAUC(): Bootstrap the AUC and conduct T-Tests for a collection of signatures.

tableAUC(): Create a table of results for t-tests and bootstrapped AUCs for multiple scored signatures.

compareBoxplots(): Create a comparison plot of boxplots for bootstrapped AUC values.

signatureROCplot(): Create an array of ROC plots to compare signatures.

signatureROCplot_CI(): Create an array of ROC plots with confidence interval bands to compare signatures.

Logistic Regression Signature Evaluation

Functions that aid in the evaluation of signatures’ predictive accuracy via logistic regression and bootstrapping procedures.

SignatureQuantitative(): Use logistic regression and bootstrap LOOCV to evaluate signatures.

plotQuantitative(): Create a boxplot using logistic regression and bootstrap LOOCV to evaluate signatures.

deseq2_norm_rle(): Normalize gene expression count data.

Bootstrap_LOOCV_LR_AUC(): Bootstrap on Leave-one-out CV with Logistic Regression.

LOOAUC_simple_multiple_noplot_one_df(): Perform Leave-one-out CV with Logistic Regression.

Example Datasets

Datasets able to be used to test functions and run examples.

TB_hiv: An example TB dataset with TB/HIV data.

TB_indian: An example TB dataset with Indian population data.

OriginalTrainingData: Discovery datasets for corresponding gene signatures.

Original Models

A collection of functions for the use of evaluating certain signatures’ performance using their original models.

.OriginalModel_NoRetraining(): TB gene signatures that do not require retraining.

.OriginalModel_Retraining(): TB gene signatures that require retraining.

ObtainSampleScore_OriginalModel(): Obtain training data, testing data, and train signature's original model.

cv_glmnet_OriginalModel(): Train original model for gene signatures Leong_24, Leong_RISK_29, Zhao_NANO_6 using lasso logistic regression.

knn_OriginalModel(): Train original model for gene signatures Berry_393 and Berry_OD_86.

lda_OriginalModel(): Train original model for gene signatures Jacobsen_3 and Sambarey_HIV_10.

randomForest_OriginalModel(): Train original model for gene signatures Maertzdorf_4, Maertzdorf_15, Verhagen_10, and LauxdaCosta_OD_3.

svm_OriginalModel(): Train original model for gene signatures Bloom_OD_144 and Zak_RISK_16.

evaluateOriginalModel(): A function that implements the original methods for multiple TB signatures.

SulimanOriginalModel(): Train original model gene signature Suliman_RISK_4.

subsetGeneSet(): Filter gene expression value matrix based on certain gene sets.

ref_combat_impute(): A function for reference batch correction and imputation.