Surprisal Analysis Guidelines

Surprisal Analysis, an R package for information theoretic analysis of gene expression data

library(SurprisalAnalysis)
library(ggplot2)

Read data and apply Surprisal analysis

data <- read.csv(system.file("extdata", "helper_T_cell_0_test.csv", package = "SurprisalAnalysis"), header=TRUE)
results <- surprisal_analysis(data)
results[[2]]-> transcript_weights
percentile_GO <- 0.95 #change based on your preference
lambda_no <- 2 #change based on your preference, lambda #1 is the baseline state
GO.results <- GO_analysis_surprisal_analysis(transcript_weights, percentile_GO, lambda_no, key_type = "SYMBOL", flip = FALSE, species.db.str =  "org.Mm.eg.db", top_GO_terms=15)
## 'select()' returned 1:1 mapping between keys and columns

The function GO_analysis_surprisal_analysis() runs Gene Ontology (GO) enrichment on the most influential transcripts from a chosen Surprisal pattern. Below are the input arguments:

  • transcript_weights
    A matrix of transcript weights, typically the second element ([[2]]) returned from the Surprisal analysis function.
  • percentile_GO
    A numeric value between 0 and 1 specifying the quantile cutoff for transcript selection. Example: 0.95 means only the top 5% of transcripts (by absolute weight) in the chosen \(\lambda\) pattern are used.
  • lambda_no
    An integer specifying which \(\lambda\) pattern to analyze. Note: \(\lambda_1\) represents the balance state, while higher-order \(\lambda\)’s capture additional constraints or patterns.
  • key_type

    The type of transcript identifiers used in your data. Options include:

    “SYMBOL” (gene symbols, e.g. TP53),

    “ENTREZID” (Entrez gene IDs),

    “ENSEMBL” (Ensembl IDs),

    “PROBEID” (microarray probe IDs). This must match the ID format in your input dataset.
  • flip
    Logical (TRUE/FALSE). If TRUE, multiplies transcript weights for the selected \(\lambda\) by –1 before selecting the top quantile. Useful for ensuring consistency with the direction of \(\lambda\) plots.
  • species.db.str

    The organism database to use for gene mapping. Current options:

    “org.Hs.eg.db” for Homo sapiens (human),

    “org.Mm.eg.db” for Mus musculus (mouse)
  • ont

    The GO ontology branch for enrichment analysis. Options:

    “BP” – Biological Process (default),

    “MF” – Molecular Function,

    “CC” – Cellular Component
  • pAdjustMethod
    The multiple testing correction method. Options include: “BH” (default), “bonferroni”, “holm”, “hochberg”, “hommel”, “BY”, “none”.
  • top_GO_terms
    An integer specifying the number of top enriched GO terms to return (default: 15).
  • ggplot(GO.results, aes(x=Description, y=Count, fill=p.adjust))+geom_bar(stat="identity")+scale_fill_gradient(low = "#790915", high = "#062c5c")+theme_minimal()+
      
      theme(
        # Remove panel border
        panel.border=element_blank(),  
        #plot.border = element_blank(),
        # Remove panel grid lines
        panel.background = element_blank(),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        # Add axis line
        axis.line = element_line(colour = "black"),
        #axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        #axis.text = element_blank(),
        #legend.position = "none",
        plot.title = element_text(hjust = 0.5, size=20),
        #axis.text = element_text(size = 15),
        
        text = element_text(size=18)
      ) +coord_flip()+labs(tag="A", title="GO analysis")