Ilesta assembles long reads (ONT or PacBio) using the Overlap-Layout-Consensus approach. It finds overlaps between reads, builds and simplifies an overlap graph, and calls a consensus sequence for the final contigs.
Install · Quick usage · How it works · Command line usage
Ilesta uses minimap2 for read alignment. The Bioconda package installs minimap2 automatically. For binary and source installations, install minimap2 separately and make sure it is available on your PATH.
conda create -n ilesta_env -c conda-forge -c bioconda ilesta
conda activate ilesta_envmkdir ilesta_workspace && cd ilesta_workspace
pixi init
pixi add -c conda-forge -c bioconda ilestamkdir Ilesta && cd Ilesta
wget https://github.com/yvlaere/Ilesta/releases/latest/download/ilesta-linux-x86_64
chmod +x ilesta-linux-x86_64
mv ilesta-linux-x86_64 Ilesta
# Optional: add Ilesta to PATH
export PATH="$PWD:$PATH"Requires Rust 1.85.0 or later.
cargo install ilestaUse a Rust toolchain that satisfies the requirements in Cargo.toml.
git clone https://github.com/yvlaere/Ilesta.git
cd Ilesta
cargo build --release
# Optional: add the binary to PATH
export PATH="$PWD/target/release:$PATH"Caution
For Ilesta versions earlier than 1.5.0, use --overhang-ratio 0.2 when running assemble. This improves assembly completeness, especially in datasets with lower coverage or shorter reads.
Ilesta assemble \
--reads-fq reads.fastq \
--output-dir out_dir \
--threads 16 \
--min-read-length 1000 \
--min-base-quality 10The read-length and mean-quality filters are optional and read filtering is disabled by default. FASTQ input may be plain or gzip-compressed. Ilesta estimates genome size from read overlaps unless --genome-size is supplied. Use --read-type to select ont (default), pb-clr or pb-hifi.
| File | Description |
|---|---|
contigs.fa |
Final assembled contigs in FASTA format. |
contigs.gfa |
Final assembly graph in GFA format. |
Intermediate files (alignments, selected reads, overlaps) are removed by default. Add --keep-intermediate to keep them.
- Prepare and align reads. Optional length and quality filters are applied, then reads are selected up to the target coverage. Ilesta estimates genome size from the overlaps unless you provide it, and uses minimap2 to find read-to-read overlaps.
- Build the overlap graph. Filtered overlaps become an oriented graph of reads. Ilesta simplifies the graph by removing transitive and redundant edges, bubbles, and tips, then compresses graph paths into draft contigs.
- Refine contigs. By default, Ilesta tries to rescue reads that are poorly represented in the draft. It removes contained contigs, uses read evidence to bridge contigs, and detects circular contigs from overlapping ends.
- Polish and write output. By default, reads are mapped back to contigs for windowed consensus polishing. The final assembly is written as
contigs.faandcontigs.gfa.
Ilesta --helpDe novo genome assembly for long reads using an overlap graph
Usage: Ilesta <COMMAND>
Commands:
align Read selection and alignment
alignment-filtering Alignment filtering
assemble Full genome assembly pipeline
scaffold Read-guided bridging of an existing contig FASTA
help Print this message or the help of the given subcommand(s)
Options:
-h, --help Print help
-V, --version Print version
Ilesta assemble --helpFull genome assembly pipeline
Usage: Ilesta assemble [OPTIONS] --reads-fq <READS_FQ>
Options:
-h, --help Print help
Output:
-p, --output-prefix <OUTPUT_PREFIX> Output prefix [default: contigs]
-o, --output-dir <OUTPUT_DIR> Output directory [default: .]
--write-dot Also write the overlap graph as graph.dot
--keep-intermediate Keep intermediate files (alignments, selected reads, overlaps) in the output directory instead of removing them; by default only the FASTA and GFA remain
Read processing and alignment:
-r, --reads-fq <READS_FQ>
Input reads in FASTQ format
--read-type <READ_TYPE>
Read type / sequencing technology [default: ont] [possible values: ont, pb-clr, pb-hifi]
-t, --threads <THREADS>
Number of threads [default: 4]
-a, --paf <PAF>
Output PAF filename [default: alignments.paf]
--min-read-length <MIN_READ_LENGTH>
Minimum read length (0 = no filtering) [default: 0]
-q, --min-base-quality <MIN_BASE_QUALITY>
Minimum average quality (0 = no filtering) [default: 0]
--genome-size <GENOME_SIZE>
Optional input genome size (if not provided, will be estimated from data)
--target-coverage <TARGET_COVERAGE>
Target read coverage for assembly input downsampling [default: 50]
--minimap2-k <MINIMAP_BATCH_SIZE>
Override minimap2's -K query minibatch size for self-alignment (e.g. 500M) [default: 500M]
--minimap2-f <MINIMAP2_F>
Override minimap2's -f high-frequency minimizer filter
--minimap2-u <MINIMAP2_U>
Override minimap2's -U minimizer occurrence bounds [default: 10,5000]
Alignment filtering:
-l, --min-overlap-length <MIN_OVERLAP_LENGTH>
Minimum overlap length [default: 2000]
-c, --min-overlap-count <MIN_OVERLAP_COUNT>
Minimum overlap count [default: 3]
-i, --min-percent-identity <MIN_PERCENT_IDENTITY>
Minimum percent identity [default: 5]
--overhang-ratio <OVERHANG_RATIO>
Overhang ratio [default: 0.2]
--overlaps <OVERLAPS>
Pre-computed overlaps binary file (optional, if provided skips alignment filtering)
Assembly:
--max-bubble-length <MAX_BUBBLE_LENGTH>
Maximum bubble length (used during bubble removal) [default: 100]
--min-support-ratio <MIN_SUPPORT_RATIO>
Minimum support ratio for bubble removal [default: 1.1]
--max-tip-len <MAX_TIP_LEN>
Maximum tip length for tip trimming [default: 4]
--fuzz <FUZZ>
Fuzz parameter for transitive edge reduction [default: 10]
--cleanup-iterations <CLEANUP_ITERATIONS>
Number of cleanup iterations to run [default: 3]
--short-edge-ratio <SHORT_EDGE_RATIO>
Short edge removal ratio (heuristic simplification) [default: 0.8]
--rescue-min-overlap-length <RESCUE_MIN_OVERLAP_LENGTH>
Minimum overlap length in the rescue stage (the unmapped reads are short and low coverage) [default: 2000]
--no-rescue
Disable rescue of poorly represented reads (enabled by default)
--seed <SEED>
Seed (the same seed gives the same result) [default: 42]
Scaffolding:
--scaffold-rounds <ROUNDS>
Contig bridging rounds (read-guided scaffolding of the final contigs; 0 disables) [default: 2]
--scaffold-min-reads <MIN_READS>
Minimum number of reads spanning a bridge between two contigs [default: 3]
--scaffold-dominance <DOMINANCE>
A competing bridge at the same contig end may have at most this fraction of the best bridge's reads [default: 0.3]
--scaffold-min-alignment-len <MIN_ALIGNMENT_LEN>
Minimum read-to-contig alignment length used for bridging [default: 1000]
--scaffold-min-identity <MIN_IDENTITY>
Minimum read-to-contig alignment identity used for bridging [default: 0.8]
Consensus:
--consensus-rounds <CONSENSUS_ROUNDS>
Consensus polishing rounds on the final contigs (0 disables) [default: 2]
--consensus-window-len <CONSENSUS_WINDOW_LEN>
Window length (bp) for consensus calling [default: 500]
--consensus-min-depth <CONSENSUS_MIN_DEPTH>
Minimum read segments per window to re-call it (else keep draft) [default: 3]
--consensus-max-depth <CONSENSUS_MAX_DEPTH>
Maximum read segments used per window [default: 25]
--no-consensus-repair
Disable replacement of stretches that no read supports but that many reads bridge
Run the default test suite with:
cargo test