10.7 Tanscriptomics Single-Omics Quick-Start Example

This section walks through a complete and standard analysis workflow for transcriptomics data, from data import to advanced functional profiling, using example datasets.

Example data download: Github link

10.7.1 Importing Transcriptomics Data

library(EasyMultiProfiler)

meta_data <- read.table('col.txt',header = T,row.names = 1)
data <- read.table('rna.txt',header = T,sep = '\t')

MAE <- EMP_easy_import(data = data,coldata = meta_data,type = 'tax')

10.7.2 Exploring Transcriptomics Data

View Current Transcriptomics Assay

MAE |>
  EMP_assay_extract() # View expression matrix

MAE |>
  EMP_coldata_extract() # View phenotype data

MAE |>
  EMP_rowdata_extract() # View gene annotations

10.7.3 Batch Effect Correction (Optional)

Correct for Batch Effects by Collection Site

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

10.7.4 Gene Identifier Conversion

The EMP package includes built-in annotation sets for Human, Mouse, Pig, and Zebrafish.

Convert SYMBOL to ENTREZID

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',species = 'Human')

Convert SYMBOL to ENSEMBL

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENSEMBL',species = 'Human')

Use a Custom OrgDb Package if Your Species is Not Built-in EMP

library(org.Hs.eg.db)
MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENSEMBL',OrgDb = org.Hs.eg.db)

10.7.5 Add Gene-Disease Associations

Add Disease Associations (Currently Supports Human_disease and Mouse_disease)

MAE |> 
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',add ='Human_disease') |> 
  EMP_assay_extract(pattern = 'cancer',pattern_ref = 'Human_disease')

10.7.6 Abundance Transformation

MAE |> 
  EMP_assay_extract() |>
  EMP_decostand(method = 'log2+1')

10.7.7 Core Gene Identification (Optional)

Identify Core Gene Set Using the edgeR Algorithm

MAE |>
  EMP_assay_extract() |>
  EMP_identify_assay(method = 'edgeR',
                     min = 10,min_ratio = 0.7)

10.7.8 Differential Expression Analysis

Perform Differential Analysis with DESeq2 and Generate Volcano Plot

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'DESeq2',.formula = ~Group) |> 
  EMP_volcanol_plot(show='pic',key_feature = c('CCL20','ZBTB7C'),
                    palette = c('#FA7F6F','#96C47D','#BEB8DC'),
                    dot_size = 2.5,threshold_x = 0.5,mytheme = "theme_light()",
                    min.segment.length = 0, seed = 42, box.padding = 0.5)

More Differential Analysis Methods with Significance Filtering

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'edgeR_quasi_likelihood',.formula = ~Group) |>
  EMP_filter(feature_condition = pvalue < 0.05,keep_result = TRUE)

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'edgeR_likelihood_ratio',.formula = ~Group) |>
  EMP_filter(feature_condition = pvalue < 0.05,keep_result = TRUE)

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'edger_robust_likelihood_ratio',.formula = ~Group) |>
  EMP_filter(feature_condition = pvalue < 0.05,keep_result = TRUE)

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'limma_voom',.formula = ~Group) |>
  EMP_filter(feature_condition = pvalue < 0.05,keep_result = TRUE)

MAE |>
  EMP_assay_extract() |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'limma_voom_sample_weights',.formula = ~Group) |>
  EMP_filter(feature_condition = pvalue < 0.05,keep_result = TRUE)

10.7.9 Machine Learning for Feature Selection

The EMP package includes Boruta, Random Forest, XGBoost, and Lasso for feature selection. For details, run help(EMP_marker_analysis).

Rapid Feature Selection with Boruta

MAE |>
  EMP_assay_extract() |>
  EMP_marker_analysis(method = 'boruta',estimate_group = 'Group') |>
  EMP_filter(feature_condition = Boruta_decision!= 'Rejected') |>
  EMP_heatmap_plot(palette='Spectral',legend_bar='auto',
                   scale='standardize',
                   clust_row=TRUE,clust_col=TRUE)

10.7.10 Over-Representation Analysis (ORA)

options(timeout = 180) # Increase timeout for database fetching

KEGG Enrichment

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',
                      species = 'Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method='DESeq2',.formula = ~Group) |>
  EMP_enrich_analysis( pvalue<0.05,keyType ='entrezid',
                       KEGG_Type = 'KEGG',species='hsa',
                      pvalueCutoff=0.05) |>
  EMP_enrich_dotplot()

GO Enrichment

library(org.Hs.eg.db)
MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'symbol',to='entrezid',species='Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'DESeq2',.formula = ~Group,
                    p.adjust = 'fdr') |> 
  EMP_enrich_analysis(pvalue<0.05,method = 'go',OrgDb=org.Hs.eg.db,
                      ont='MF',readable=TRUE,pvalueCutoff=0.05) |>
  EMP_enrich_dotplot(show=6)

DOSE Enrichment

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'symbol',to='entrezid',
                      species='Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'DESeq2',
                    .formula = ~Group,p.adjust = 'fdr') |> 
  EMP_enrich_analysis(pvalue<0.05,method = 'do',ont="HDO",
                      organism= 'hsa',readable=TRUE,
                      pvalueCutoff=0.05) |>
  EMP_enrich_dotplot(show=5)

Reactome Enrichment

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'symbol',to='entrezid',
                      species='Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method = 'DESeq2',.formula = ~Group,
                    p.adjust = 'fdr') |> 
  EMP_enrich_analysis(pvalue<0.05,method = 'Reactome',organism= 'human',
                      readable=TRUE,pvalueCutoff=0.05) |>
  EMP_enrich_dotplot()

10.7.11 GSEA Enrichment

options(timeout = 180) # Increase timeout for database fetching

Three ranking methods are available for GSEA. The examples below demonstrate KEGG enrichment.

You can apply the same parameter same from ORA to perform GSEA with GO, DOSE, and Reactome.

KEGG GSEA with Signal2Noise Ranking

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',species = 'Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_GSEA_analysis(pvalue<0.05,method='signal2Noise',
                    estimate_group = 'Group',species = 'hsa',
                    pvalueCutoff = 1,keyType = 'entrezid') |>
  EMP_GSEA_plot(geneSetID='hsa04930')

KEGG GSEA with Log2FC Ranking

MAE |>
  EMP_assay_extract() |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',
                      species = 'Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_diff_analysis(method='DESeq2',.formula = ~0+Group,
                    group_level=c('PMC','PMS1')) |>
  EMP_GSEA_analysis(method='log2FC',enrich_method = 'kegg',
                    species = 'hsa',keyType = 'entrezid',
                    pvalueCutoff = 0.05) |>
  EMP_GSEA_plot(geneSetID='hsa04950')

KEGG GSEA with Correlation-Based Ranking

MAE |>
  EMP_assay_extract() |> 
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',
                      species = 'Human') |>
  EMP_filter(Group %in% c('PMC','PMS1')) |>
  EMP_GSEA_analysis(method='cor',enrich_method = 'kegg',
                    keyType='entrezid',estimate_group = 'NR5A2',
                    cor_method = 'spearman',
                    pvalueCutoff = 0.05,species='hsa') |>
  EMP_GSEA_plot(geneSetID='hsa05415')

10.7.12 WGCNA Analysis

Step 1: Cluster Analysis Based on Phenotype

MAE |>
  EMP_assay_extract()  |> 
  EMP_identify_assay(method = 'edgeR',estimate_group = 'Group') |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',species = 'Human') |>
  EMP_WGCNA_cluster_analysis()

Step 2: Generate Heatmap of Phenotype-Correlated Gene Modules

MAE |>
  EMP_assay_extract()  |> 
  EMP_identify_assay(method = 'edgeR',estimate_group = 'Group') |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',species = 'Human') |>
  EMP_WGCNA_cluster_analysis()  |>
  EMP_WGCNA_cor_analysis(coldata_to_assay = c('NR5A2','HNF4G','HNF1B','PAX4','RFX6','NEUROG3'),
                         method='spearman') |>
  EMP_heatmap_plot(palette = 'Spectral')

Step 3: Enrichment Analysis for Selected Gene Modules

MAE |>
  EMP_assay_extract()  |> 
  EMP_identify_assay(method = 'edgeR',estimate_group = 'Group') |>
  EMP_feature_convert(from = 'SYMBOL',to = 'ENTREZID',species = 'Human') |>
  EMP_WGCNA_cluster_analysis()  |>
  EMP_WGCNA_cor_analysis(coldata_to_assay = c('NR5A2','HNF4G','HNF1B','PAX4','RFX6','NEUROG3'),
                         method='spearman') |>
  EMP_heatmap_plot(palette = 'Spectral') |>
  EMP_filter(feature_condition = WGCNA_color == 'blue' ) |> 
  EMP_enrich_analysis(keyType = 'entrezid',species = 'hsa') |>
  EMP_enrich_dotplot()

10.7 Trans example