Overview
Genome assembly is the computational process of reconstructing entire genome sequences from the short DNA fragments produced by high-throughput sequencing platforms. Because sequencing machines read only hundreds of base pairs at a time, bioinformaticians must piece together millions or billions of these reads, much like solving an enormous jigsaw puzzle. The accuracy of an assembled genome directly affects every downstream analysis, from gene prediction to comparative genomics. Modern assemblers handle the complexities of repetitive regions, sequencing errors, and varying coverage depths using sophisticated graph-based algorithms.
Key Concepts
Two main strategies exist for genome assembly. De novo assembly constructs a genome without any prior reference, relying on overlap-layout-consensus (OLC) or de Bruijn graph approaches to merge reads into contiguous sequences called contigs. Reference-guided assembly maps reads to a known reference genome and then assembles the unmapped portions, which is particularly useful for resequencing projects. Key quality metrics include N50 (the contig length at which 50% of the assembly is contained) and total assembly size. Assembly validation often involves checking against known sequences or using long-read technologies for gap closure.
Applications
Genome assembly is foundational to nearly every genomics application. It enables the discovery of novel genes, the identification of structural variants, and the characterization of non-coding regulatory elements. In medicine, assembled genomes from pathogens allow rapid outbreak tracking and antibiotic resistance profiling. Agricultural genomics relies on high-quality assemblies to map traits of economic importance. Modern projects frequently combine next-generation sequencing data with long reads and optical mapping to produce chromosome-level assemblies, building on classic DNA sequencing methods. Assembly also underpins functional studies such as CRISPR-Cas9 target design, where off-target predictions depend on an accurate reference.
Practical Protocol
For Illumina-based de novo assembly of a bacterial genome, start with quality trimming: fastp -i R1.fastq -I R2.fastq -o trimmed_R1.fastq -O trimmed_R2.fastq -q 20 -l 50. Assemble with SPAdes using multiple k-mer sizes: spades.py -1 trimmed_R1.fastq -2 trimmed_R2.fastq -o spades_output -k 21,33,55,77 --careful. The --careful flag reduces mismatches and indels by running MismatchCorrector. The output includes contigs.fasta and scaffolds.fasta. For long-read assembly with Oxford Nanopore data, use Flye: flye --nano-raw reads.fastq -o flye_out -t 8 -g 5m, where -g provides the estimated genome size. For hybrid assembly combining short and long reads, use Unicycler: unicycler -1 short_R1.fastq -2 short_R2.fastq -l long_reads.fastq -o hybrid_assembly. Assess assembly quality with QUAST: quast.py contigs.fasta -r reference.fasta -g genes.gff -o quast_output. Key metrics include N50 (contig length where 50% of assembly is covered), total assembly size, number of contigs, genome fraction (fraction of reference covered), and the number of misassemblies. For a complete bacterial genome, aim for N50 equal to the chromosome length and GC content matching the reference. Check completeness with BUSCO against the appropriate lineage: busco -i contigs.fasta -l bacteria_odb10 -o busco_out -m genome. Polish assemblies with Pilon to correct errors using read alignments: pilon --genome contigs.fasta --frags aligned.bam --output polished. For chromosome-level scaffolding, use Hi-C data with SALSA2 or 3D-DNA to order and orient contigs into chromosome-scale pseudomolecules.
