cpg

CpG island analysis toolkit written in Rust. Accepts FASTA input and provides three subcommands: island detection, sequence statistics, and sliding-window profiling.

Install

Download the binary for your system from the latest release:

Linux

curl -LO https://github.com/marceelrf/cpg/releases/latest/download/cpg-linux-x86_64
chmod +x cpg-linux-x86_64
mv cpg-linux-x86_64 ~/.local/bin/cpg

macOS

curl -LO https://github.com/marceelrf/cpg/releases/latest/download/cpg-macos-aarch64
chmod +x cpg-macos-aarch64
mv cpg-macos-aarch64 ~/.local/bin/cpg

Windows — download cpg-windows-x86_64.exe from the releases page.

Subcommands

finder — detect CpG islands

# Built-in preset (default: Gardiner-Garden & Frommer 1987)
# Default: Gardiner-Garden & Frommer 1987
# The thread number default is 4
cpg finder -i genome.fa --threads 3

# Takai & Jones 2002
cpg finder -i genome.fa --criteria takai

# Fully custom thresholds
cpg finder -i genome.fa --min-length 300 --min-gc 0.52 --min-obs-exp 0.62

# Mix: start from a preset and override one threshold
cpg finder -i genome.fa --criteria takai --min-length 300

# BED output — open directly in IGV or bedtools
cpg finder -i genome.fa --output bed > islands.bed

Built-in presets:

Preset	Length	GC%	Obs/Exp
gardiner	≥ 200	≥ 50	≥ 0.60
takai	≥ 500	≥ 55	≥ 0.65

stats — per-sequence statistics

cpg stats -i sequences.fa
cpg stats -i sequences.fa --output tsv
cpg stats -i sequences.fa --output json

Outputs: ID, Length, CG_count, GC%, Obs/Exp.

profiler — sliding-window metric profile

# Obs/Exp in 200 bp windows stepping 50 bp (default)
cpg profiler -i genome.fa --window 200 --step 50 --metric obs_exp

# GC% — TSV output, easy to pipe into R or Python
cpg profiler -i genome.fa -w 100 -s 10 -m gc_percent --output tsv

# BEDGRAPH output — load directly into IGV or UCSC Genome Browser
cpg profiler -i genome.fa -w 200 -s 50 -m obs_exp --output bedgraph > profile.bedgraph

Metrics: cg_count, gc_percent, obs_exp.

Output formats

Format	Available in	Use case
table	all	terminal / quick inspection
tsv	all	R, Python, Excel
json	all	downstream scripts
bed	finder	IGV, bedtools, UCSC
bedgraph	profiler	IGV, UCSC Genome Browser

All subcommands accept --output as a global flag.

Funding

We thank FAPESP(2025/28063-3) and CAPES(PIPD 88887.114278/2025-00) for the financial support.