12. De Novo Genome & Transcriptome Assembly

SPAdes is the benchmark de novo assembler for bacterial, fungal, and viral genomes. It employs iterative multi-sized de Bruijn graphs and read error correction to produce contigs and scaffold graphs, with dedicated modes for plasmids and metagenomes.

# 1. Run SPAdes bacterial genome assembly with multi-k-mer iteration
spades.py \
  -1 sample_R1.fastq.gz -2 sample_R2.fastq.gz \
  -k 21,33,55,77,99,127 \
  --careful \
  --threads 16 \
  --memory 64 \
  -o spades_isolate_out/

# Output contigs: spades_isolate_out/contigs.fasta
# Output scaffolds: spades_isolate_out/scaffolds.fasta

MEGAHIT assembles complex, large-scale metagenomic datasets using succinct de Bruijn graphs. It consumes minimal memory and scales smoothly to terabyte-scale FASTQ archives without requiring massive supercomputing nodes.

# 1. Assemble metagenomic paired-end reads with succinct de Bruijn graph
megahit \
  -1 sample_R1.fq.gz -2 sample_R2.fq.gz \
  --k-list 21,29,39,59,79,99,119,141 \
  --min-contig-len 500 \
  -m 0.85 \
  -t 24 \
  -o megahit_assembly/

# Final contigs: megahit_assembly/final.contigs.fa

Flye is a de novo assembler for single-molecule sequencing reads using repeat graphs. It resolves complex structural genomic repeats, assembling complete circular bacterial chromosomes and eukaryotic chromosome arms from ONT and PacBio data.

# 1. Assemble Oxford Nanopore long reads
flye \
  --nano-hq ont_reads.fq.gz \
  --genome-size 5m \
  --out-dir flye_assembly/ \
  --threads 16 \
  --iterations 2

# Output assembly: flye_assembly/assembly.fasta
# Assembly graph (GFA): flye_assembly/assembly_graph.gfa

Hifiasm produces haplotype-resolved, chromosome-level de novo genome assemblies from PacBio HiFi reads. When provided with Hi-C or parental trio data, it phases both maternal and paternal haplotypes end-to-end without reference bias.

# 1. Haplotype-resolved assembly with PacBio HiFi reads and Hi-C pairs
hifiasm \
  -o asm_sample \
  -t 24 \
  --h1 hic_R1.fq.gz --h2 hic_R2.fq.gz \
  pacbio_hifi.fastq.gz

# 2. Extract phased FASTA contigs from GFA graphs
awk '/^S/{print ">"$2"\n"$3}' asm_sample.hic.hap1.p_ctg.gfa > hap1_contigs.fasta
awk '/^S/{print ">"$2"\n"$3}' asm_sample.hic.hap2.p_ctg.gfa > hap2_contigs.fasta

Canu is an Overlap-Layout-Consensus assembler designed for noisy single-molecule reads. It executes automated read self-correction with MinHash alignment (MHAP), read trimming, and unitig generation.

# 1. Run end-to-end Canu assembly on Oxford Nanopore reads
canu \
  -p sample_asm -d canu_out/ \
  genomeSize=4.8m \
  maxThreads=16 \
  -nanopore ont_reads.fq.gz

Medaka is a consensus tool developed by Oxford Nanopore. It aligns ONT reads to draft assemblies and uses recurrent neural networks to resolve remaining insertion/deletion errors and homopolymer run ambiguities.

# 1. Polish draft assembly with chemistry-specific neural network model
medaka_consensus \
  -i ont_reads.fq.gz \
  -d draft_assembly.fasta \
  -o medaka_polished/ \
  -t 16 \
  -m r1041_e82_400bps_sup_v4.2.0

# Polished consensus in: medaka_polished/consensus.fasta

Racon rapidly computes consensus sequences using SIMD-accelerated partial order alignment. Pilon then aligns high-accuracy paired-end Illumina reads to the draft assembly to fix single-base substitution errors and indels.

# 1. Fast long-read consensus polishing with Racon
minimap2 -t 16 draft.fasta ont_reads.fq.gz > ont_to_draft.paf
racon -t 16 ont_reads.fq.gz ont_to_draft.paf draft.fasta > racon_polished.fasta

# 2. Map high-accuracy short reads to consensus
bwa index racon_polished.fasta
bwa mem -t 16 racon_polished.fasta illumina_R1.fq.gz illumina_R2.fq.gz \
  | samtools sort -@ 4 -o illumina_to_draft.sorted.bam -
samtools index illumina_to_draft.sorted.bam

# 3. Final base-error polishing with Pilon
pilon --genome racon_polished.fasta \
  --frags illumina_to_draft.sorted.bam \
  --output pilon_final_polished \
  --vcf --changes