Home / Glossary

Glossary

Concise definitions for terms used in the course so far. This list will grow as later modules introduce new concepts.

Absolute path

A file or directory path written from the root of the filesystem (starting with /). Unambiguous regardless of your current working directory.

Accession

A unique identifier assigned to a database record — for example, a specific genome assembly or sequencing run — so it can be unambiguously cited and retrieved.

Adapter

A short, known synthetic DNA sequence ligated onto fragments during library preparation so they can bind to the sequencer and be identified. Not part of the biological sample itself.

Adapter read-through

When a DNA fragment is shorter than the sequencing read length, the read continues past the fragment's end into adapter sequence. Common with short, degraded (e.g. ancient/historical) DNA, and detected by FastQC's Adapter Content panel.

BioSample

The NCBI/ENA database record (accession prefix SAMN or SAMEA) describing a biological sample or specimen used for sequencing. One BioSample can be linked to several sequencing libraries and runs — it describes the source, not any single sequencing experiment.

BUSCO

Benchmarking Universal Single-Copy Orthologs — a tool and completeness metric reporting what proportion of a fixed set of genes expected in nearly every member of a lineage are found intact in a given assembly. A proxy for how much real, unbroken biology the assembly actually contains.

Chromosome-level assembly

A genome assembly in which sequence has been ordered and oriented into complete chromosomes, rather than left as many shorter, unordered fragments.

Compressed file

A file that has been algorithmically shrunk to take up less disk space. Must be decompressed (or read with compression-aware tools) to access its original content. Gzip (.gz) is the most common format for genomic data.

Compute node

On an HPC cluster such as Mjolnir, a machine dedicated to running scheduled jobs (via Slurm) — where actual analyses, alignments, and other resource-intensive work should run, as opposed to a login node.

Contig

A contiguous stretch of assembled sequence built by overlapping sequencing reads, with no gaps. Short for "contiguous sequence."

Duplication / library complexity

The proportion of sequencing reads that are identical (or near-identical) to each other. High duplication can mean a sample was sequenced far deeper than its library's actual molecular diversity ("complexity") supports — a common issue with low-input material such as historical or ancient DNA, not only a sign of over-sequencing.

ENA

The European Nucleotide Archive — EMBL-EBI's database of nucleotide sequencing data, including raw reads, assemblies, and their associated metadata. Mirrors much of the same data held by NCBI's SRA.

FASTA

A plain-text format for representing one or more sequences. Each sequence begins with a header line starting with >, followed by the sequence itself on the following line(s).

FASTQ

A plain-text format for representing sequencing reads together with per-base quality scores. Each read is exactly four lines: an identifier line starting with @, the called sequence, a separator line (+), and a quality string the same length as the sequence. Almost always distributed as .fastq.gz.

FastQC

A diagnostic tool that reads a FASTQ file and reports summary statistics and quality patterns (per-base quality, GC content, duplication, adapter content, and more) as an HTML report. It never modifies the file it analyses, and its pass/warn/fail flags are heuristics calibrated on typical modern libraries — not an automatic verdict on whether a dataset is usable.

GC content

The proportion of a sequence's bases that are G or C rather than A or T. FastQC reports this per read; an unusual or shifted distribution can have several possible causes (sample composition, contamination, bias), not one single explanation.

GCA_ / GCF_ accession prefix

NCBI genome assembly accessions starting GCA_ are GenBank assemblies (submitted by any group, not independently curated by NCBI); those starting GCF_ are RefSeq assemblies, which NCBI has additionally processed and annotated. The prefix reflects database status, not assembly quality.

Genome assembly

A reconstruction of an organism's genome built computationally from sequencing reads, consisting of contigs and/or scaffolds (and, ideally, full chromosomes).

Heterospecific mapping

Aligning sequencing reads from one species against a reference genome from a different, related species — necessary when no reference exists for the species itself (as with the extinct Great Auk), but prone to reference bias: increasing sequence divergence causes more mismatches, reduced mapping of divergent regions, and under-detection of true heterozygosity.

Login node

On an HPC cluster such as Mjolnir, the shared machine you land on when you connect — meant for lightweight tasks like file organisation, script editing, and job submission, not for running analyses directly.

Metadata

Data about data — for a genome assembly, this includes things like the submitting organisation, sequencing technology used, assembly date, and quality statistics, rather than the sequence itself.

MultiQC

A tool that scans a directory of FastQC (and other tool) outputs and aggregates them into a single comparative HTML report — reading the parsed .zip/log files each tool produces, not their HTML reports.

N50 (contig / scaffold)

A contiguity statistic: the length at which half of an assembly's total size is contained in contigs (or scaffolds) at least that long. A higher N50 generally means a more contiguous assembly, but contig N50 and scaffold N50 measure different things and should not be compared to each other directly.

Paired-end sequencing

A sequencing design where each DNA fragment is read from both ends, producing two files per run — R1 and R2. R1 and R2 are mates describing the same physical fragment, not independent datasets, and their read order must stay synchronized between the two files.

Phred quality score

A per-base confidence value assigned during base calling, encoded as a single ASCII character in a FASTQ quality line. Higher values indicate the sequencer was more confident in that base call.

Project directory (Mjolnir)

On Mjolnir, the shared location /projects/<project>/ where a research project's data lives, with a standard internal layout including a people/<KU-ID>/ folder for each member's own working files — distinct from your personal, size-limited home directory.

R1 / R2

The conventional names for the two FASTQ files produced by a paired-end sequencing run — R1 from one end of each fragment, R2 from the other. Read n in R1 and read n in R2 are mates and must stay in the same order across both files.

Read

A single sequence produced by a sequencer for one DNA fragment (or one end of it), represented as one four-line record in a FASTQ file.

Reference bias

Systematic distortion introduced when sequencing reads are mapped to a reference genome that differs from the sampled individual's own genome — reads or alleles that resemble the reference are retained preferentially, which can under-detect true heterozygosity. Increases with evolutionary distance between sample and reference.

Reference genome

A genome assembly chosen to serve as the standard coordinate system that other data (reads, variants, annotations) are compared or aligned against.

Relative path

A file or directory path interpreted relative to your current working directory. Convenient, but only correct if you know where you're starting from.

Run accession

The ENA/SRA accession (prefix SRR or ERR) identifying one actual execution of sequencing on a library, and the FASTQ file(s) it produced. A library's experiment accession (SRX/ERX) can in principle have more than one run accession under it.

Scaffold

A sequence built by ordering and orienting multiple contigs, with gaps of estimated length (often represented as runs of N) between them where the exact sequence is unknown.

Sequencing library

A prepared, adapter-ligated collection of DNA molecules ready for sequencing, built from a sample. A single specimen can yield several distinct libraries (different extractions, protocols, or sequencing batches) — more libraries does not automatically mean more individuals were sampled.

Single-end sequencing

A sequencing design where each fragment is read from only one end, producing a single FASTQ file per run. Common for ancient DNA, where fragments are often already shorter than a single read.