visualisation

Visualisation

Efesto includes scripts/plot_heatmap.R for publication-ready heatmaps.

---

Features

• Hierarchical clustering (Ward's method) on both axes
• White → orange → red for gene counts; white → blue for coverage
• Static PDF/PNG via pheatmap
• Interactive self-contained HTML via plotly
• Compatible with FeGenie's R script input format

---

Installation

conda install -c conda-forge r-pheatmap r-plotly r-htmlwidgets r-optparse r-rcolorbrewer r-scales

---

Usage

# Process entire results directory (auto-detects CSV files)
Rscript scripts/plot_heatmap.R --input results/

# Static PDF only
Rscript scripts/plot_heatmap.R --input results/ --type static --format pdf --out figures/

# Interactive HTML, filter low-count categories
Rscript scripts/plot_heatmap.R --input results/ --type interactive --min_count 1

# Specify a single CSV
Rscript scripts/plot_heatmap.R --input results/Efesto-heatmap-data.csv

---

Arguments

Argument	Default	Description
--input	(required)	CSV file or output directory
--type	both	static | interactive | both
--format	both	pdf | png | both (static only)
--out	same as input	Output directory
--width / --height	auto	Plot dimensions in inches
--min_count	0	Minimum value to include a category row
--no_cluster_rows	off	Disable row clustering
--no_cluster_cols	off	Disable column clustering

---

Input files

The script reads:

• Efesto-heatmap-data.csv — gene-count matrix (categories × genomes)
• Efesto-coverage-heatmap.csv — coverage matrix (same shape; optional)

Both are in FeGenie's original CSV format and are compatible with FeGenie's own R visualisation script.

---

Custom downstream analysis

For custom plots, Efesto-results-long.tsv is the recommended starting point (one row per ORF, all columns including cluster_confidence and model_nseq):

library(tidyverse)

hits <- read_tsv("results/Efesto-results-long.tsv")

# Mean confidence by category
hits |>
  group_by(category) |>
  summarise(mean_conf = mean(cluster_confidence, na.rm=TRUE), n = n()) |>
  arrange(desc(mean_conf))

# Iron reduction genes with high confidence
hits |>
  filter(str_detect(category, "iron_reduction"), cluster_confidence >= 0.8) |>
  select(genome, contig, gene, bitscore, cluster_confidence)