Genomic & Bioinformatics File Formats

FASTA is the simplest and most universally used format for storing biological sequences. Named for the FASTA sequence alignment program (1985), it stores one or more sequences with associated metadata in a plain text format that is both human-readable and machine-parsable.

Format Structure

Each sequence entry consists of:

1. Header line: Begins with > (greater-than symbol), followed by a sequence identifier and optional description
2. Sequence data: One or more lines of sequence characters. Lines are typically 60–80 characters wide by convention.

Example — Multiple Sequences

Nucleotide IUPAC Codes

Common Uses

• Reference genome sequences (human GRCh38, mouse GRCm39)
• Protein sequence databases (UniProt, NCBI RefSeq proteins)
• Transcript sequences (GENCODE, Ensembl)
• BLAST and sequence alignment queries
• Primer sequences, probe sequences

FASTQ extends FASTA by including per-base quality scores alongside the sequence. It is the standard output format of next-generation sequencing (NGS) instruments — Illumina, Oxford Nanopore, PacBio platforms all produce FASTQ output. Every short read sequencing experiment begins with FASTQ files.

Format Structure

Each read is represented by exactly 4 lines:

1. Line 1 — Identifier: Begins with @, followed by sequence identifier and optional description (instrument, flowcell, coordinates)
2. Line 2 — Sequence: Raw nucleotide sequence
3. Line 3 — Separator: Always + (optionally followed by the identifier again)
4. Line 4 — Quality scores: ASCII-encoded Phred quality scores, one character per base

Example

Phred Quality Scores

Quality scores (Q-scores) quantify the probability of a base call error. The relationship is: Q = -10 × log₁₀(P), where P is the error probability.

Modern Illumina sequencing typically achieves Q30+ for 75–90% of bases. Reads are typically quality-filtered (remove low-quality bases/reads) and adapter-trimmed before downstream analysis using tools like Trimmomatic, fastp, or Cutadapt.

Paired-End Sequencing

Most Illumina sequencing is paired-end: each fragment is sequenced from both ends, producing two FASTQ files (R1 and R2) where reads at the same position in each file are paired. Paired reads are more informative for alignment and structural variant detection.

VCF (Variant Call Format) is the standard format for storing genetic variants identified from sequencing data — single nucleotide polymorphisms (SNPs), insertions/deletions (indels), and structural variants. It is produced by variant calling tools such as GATK HaplotypeCaller, DeepVariant, and Strelka2, and used by major variant databases (dbSNP, ClinVar, gnomAD).

Format Structure

Core VCF Columns

Genotype Encoding (GT field)

• 0/0 — Homozygous reference (no variant)
• 0/1 — Heterozygous (one reference, one alternate allele)
• 1/1 — Homozygous alternate
• 0|1 — Phased heterozygous (paternal|maternal allele)
• ./. — Missing data (insufficient coverage)

BED (Browser Extensible Data) format is the standard way to represent genomic intervals — specific chromosomal regions defined by start and end coordinates. It is used for gene annotations, ChIP-seq peaks, ATAC-seq open chromatin regions, CpG islands, repeat elements, regulatory regions, and virtually any genomic feature defined by position.

BED Core Fields

SAM (Sequence Alignment/Map) is a text-based format for storing read alignments — the results of mapping short sequencing reads to a reference genome using tools like BWA-MEM, HISAT2, STAR, or Bowtie2. BAM is the binary compressed equivalent of SAM, dramatically smaller in size and much faster to parse. BAM files are the primary storage format for aligned sequencing data in production pipelines.

SAM File Structure

CIGAR String — Alignment Encoding

The CIGAR (Compact Idiosyncratic Gapped Alignment Report) string describes the alignment of a read to the reference:

BAM vs SAM vs CRAM

Format Conversion Tools

Format Comparison Summary