Assembling the Code: Genome Sequencing Quiz

Whole genome shotgun sequencing works by randomly shearing the entire genome into millions of small overlapping fragments. Each fragment is sequenced individually, producing short sequence reads. Powerful computational algorithms then scan all reads for overlapping sequences and assemble them into longer contiguous sequences called contigs, which are further joined into larger scaffolds. This strategy allows rapid sequencing of large genomes without the need to first physically map or order the fragments before sequencing.

Modern next-generation sequencing platforms do not require cloning of fragments into bacterial vectors before sequencing. Early shotgun approaches using Sanger sequencing did require cloning into vectors such as plasmids or cosmids. Current platforms such as Illumina sequencing prepare libraries by ligating synthetic adapters directly onto fragmented DNA, then amplify and sequence fragments directly without biological cloning steps. This shift eliminated the time-consuming and bias-introducing cloning steps required in earlier whole genome shotgun methodologies.

A sequencing read is the raw nucleotide sequence generated from a single DNA fragment during a sequencing run. Read lengths vary by platform, ranging from approximately 75 to 300 base pairs for short-read platforms such as Illumina to tens of thousands of base pairs for long-read platforms such as Pacific Biosciences and Oxford Nanopore. The vast numbers of reads produced, often billions per run, are then computationally assembled by identifying overlapping sequences to reconstruct the original genome.

A contig is a contiguous assembled sequence produced by joining overlapping reads from a sequencing run. Contigs represent regions of the genome that have been assembled without interruption into a single unbroken sequence. In practice, genome assembly produces many contigs because repetitive sequences and sequencing gaps prevent complete assembly of chromosomes in one step. Contigs are subsequently linked into scaffolds using paired-end read information, and scaffolds are ordered into pseudochromosomes using additional mapping data.

Repetitive sequences are a major challenge because reads originating from different copies of a repeat are indistinguishable, confusing the assembler. Diploid heterozygosity produces two versions of each genomic region that must be correctly separated during assembly. Short reads cannot span long repeats, creating gaps and misassemblies. Sequencing errors are not automatically corrected by the instrument and must be addressed computationally during base calling and assembly quality filtering steps.

Sequencing coverage depth is calculated as the total number of base pairs in all reads divided by the estimated genome size. Higher coverage means each position in the genome is represented by more independent reads, which helps distinguish true sequence from sequencing errors, resolve repetitive regions, and detect heterozygous variants. Most whole genome sequencing projects aim for a minimum coverage of 30-fold for reliable variant detection, with higher coverage used for structural variant analysis and genome assembly projects.

Paired-end sequencing reads both ends of a DNA fragment of known approximate length. This produces two reads with a predictable distance and relative orientation between them. During assembly, pairs whose individual reads map to different contigs provide evidence that those contigs are adjacent in the genome, allowing assemblers to link them into larger scaffolds even across regions that cannot be directly sequenced. Paired-end data is especially valuable for spanning repetitive regions and detecting structural genomic variants.

Long-read sequencing platforms generate reads ranging from tens of thousands to over one million base pairs in length. These long reads can span repetitive elements and structural variants that are impossible to resolve with short reads of 150 to 300 base pairs. This results in genome assemblies with far fewer and larger contigs, approaching chromosome-level completeness in many organisms. The main trade-off is that long-read platforms historically have had higher per-base error rates than short-read platforms, though accuracy has improved substantially.

N50 is a standard genome assembly quality metric. To calculate it, all contig lengths are sorted from longest to shortest, and a running sum is computed until half the total assembly size is reached. The length of the contig at that point is the N50 value. A higher N50 indicates a more contiguous assembly with fewer, longer contigs, which generally means the assembly better represents the true chromosome structure. N50 is reported alongside total assembly size and number of contigs to provide a comprehensive picture of assembly quality.

A scaffold is produced during genome assembly by using paired-end or mate-pair read information to order and orient multiple contigs relative to one another. The regions between ordered contigs within a scaffold are represented by runs of the letter N, indicating gaps of estimated but unsequenced sequence. Scaffolds are therefore larger than individual contigs but may contain internal gaps. Further anchoring of scaffolds to a physical or genetic map, or using chromosome conformation capture data, allows scaffolds to be assigned to specific chromosomes.

The Human Genome Project primarily used a hierarchical clone-by-clone approach combined with shotgun sequencing of individual bacterial artificial chromosome clones. Subsequent efforts used whole genome shotgun sequencing for speed. Paired-end and mate-pair libraries improved scaffold construction and gap closing. More recent reference genome improvements have incorporated long-read platforms to close remaining gaps and resolve previously inaccessible repetitive regions, resulting in the first complete telomere-to-telomere human genome assembly published in 2022.

Repetitive sequences are the primary barrier to complete genome assembly using short-read shotgun approaches. When reads originating from different copies of a repeat element are indistinguishable from one another, the assembler cannot determine how many copies are present or in what order they are arranged. This collapses repeats into single incorrectly assembled regions or leaves gaps. Long-read platforms that generate reads longer than the repeat unit can span entire repetitive arrays, resolving their structure and enabling complete chromosome assemblies.

De novo assembly reconstructs a genome sequence from scratch without using any previously sequenced reference genome, using only overlap information between reads. It is used when sequencing a new organism for the first time. Reference-guided assembly maps reads from a new individual or strain against an existing reference genome of a closely related organism to identify variants and fill gaps. Both strategies are valuable, and many projects use both in combination depending on the goals of the sequencing effort and the availability of existing reference sequences.

Base calling is the process by which sequencing instrument software interprets the raw signal produced during sequencing, whether fluorescence in Illumina or ionic current in Nanopore, and assigns a nucleotide identity to each position. The confidence in each base call is expressed as a Phred quality score, where a score of 30 means there is a one-in-one-thousand probability of an error. Phred scores are used to filter low-quality reads and trim poor-quality ends before genome assembly or variant calling.

A standard whole genome shotgun workflow involves fragmenting genomic DNA into size-selected pieces, ligating sequencing adapters to create libraries, loading libraries onto the sequencing platform to generate millions or billions of reads, and then using computational tools such as genome assemblers and read mappers to reconstruct the original genome sequence. Manual review of individual reads before assembly is not part of standard workflow; automated quality filtering and bioinformatic pipelines handle all computational processing steps.