Genomics File Formats Reference

SAM / BAM

SAM (Sequence Alignment/Map) is a tab-delimited text format of read alignments against a reference; BAM is its compressed binary form. Each alignment line has 11 mandatory fields. POS is 1-based.

SAM FLAG bits

The FLAG field is the sum of the set bits below — for example 99 = 1 + 2 + 32 + 64 (paired, proper pair, mate reverse strand, first in pair).

VCF

VCF (Variant Call Format) stores variant calls. A file begins with ## meta header lines, then a single #CHROM column-header line, then one record per variant. POS is 1-based.

The first eight columns are fixed; FORMAT and one column per sample appear only when genotypes are recorded.

##fileformat=VCFv4.3
#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	sample1
chr1	1000	.	A	G	60	PASS	DP=42	GT	0/1

BED

BED is a simple interval format. It is 0-based, half-open. Only the first three columns are required (BED3); BED6 adds name, score and strand, and BED12 adds the block columns that describe exon structure.

BED3 = columns 1–3 · BED6 = columns 1–6 · BED12 = all 12 columns.

GFF3 / GTF

GFF3 and GTF are 9-column, tab-delimited annotation formats. Both are 1-based and closed. GTF, an extension of GFF2, shares the first eight columns with GFF3 but differs in column 9, the attributes.

The difference is entirely in column 9. GFF3 uses key=value;key=value pairs (with ID= and Parent= to link features):

chr1	ensembl	gene	1000	2000	.	+	.	ID=gene1;Name=EXMP

GTF uses key "value"; pairs (notably gene_id and transcript_id):

chr1	ensembl	CDS	1000	2000	.	+	0	gene_id "gene1"; transcript_id "tx1";

See also

• Sequence File Formats — FASTA, FASTQ and GenBank →
• Phred Quality Scores — error probabilities and ASCII encodings →
• Understanding FASTQ Format, Phred Quality Scores and N50 →

Sources

1. 1
   Sequence Alignment/Map Format Specification (version b5341fb, 12 Aug 2025)

   The SAM/BAM Format Specification Working Group (samtools/hts-specs) · 2025

   The entire SAM_FIELDS table (all 11 mandatory fields, columns 1-11, QNAME through QUAL, in exactly the page's order) and the entire SAM_FLAGS table (all 12 rows, 0x1/1 through 0x800/2048, with the spec's wording mapping one-to-one onto the page's plain-English meanings: 0x4 'segment unmapped', 0x10 'SEQ being reverse complemented', 0x100 'secondary alignment', 0x200 'not passing filters, such as platform/vendor quality controls', 0x400 'PCR or optical duplicate', 0x800 'supplementary alignment'). Also backs the COORDS row 'SAM / BAM — 1-based — fully closed', the 'POS is 1-based' claims in the SAM section, and FAQ item 1's '11 mandatory fields'. The FLAG-as-sum-of-bits rule and the 99 = 1+2+32+64 worked example follow directly from the bit table. Verified by extracting the PDF text locally. The peer-reviewed companion paper is Li H et al., 'The Sequence Alignment/Map format and SAMtools', Bioinformatics 2009;25(16):2078-9, doi:10.1093/bioinformatics/btp352, PMID 19505943 — cite the spec for the tables, the paper for the format's provenance.
2. 2
   The Variant Call Format Specification, VCFv4.3 and BCFv2.2 (version c101c79, 5 Nov 2025)

   samtools/hts-specs · 2025

   The whole VCF_COLUMNS table and the VCF section prose. Section 1.6.1: "There are 8 fixed fields per record" — CHROM, POS, ID, REF, ALT, QUAL, FILTER, INFO, matching the page's columns 1-8 including 'QUAL — Phred-scaled variant quality' (spec: "Phred-scaled quality score for the assertion made in ALT. i.e. -10log10 prob(call in ALT is wrong)") and 'FILTER — PASS or filters the variant failed'. Section 1.5 backs column 9: "If genotype data is present in the file, these are followed by a FORMAT column header, then an arbitrary number of sample IDs", which is exactly the page's 'the first eight columns are fixed; FORMAT and one column per sample appear only when genotypes are recorded'. POS 1-based is verbatim: "The reference position, with the 1st base having position 1" — backing the COORDS row, the section prose and FAQ item 5. Also backs the '##' meta lines / single '#CHROM' header line structure and the VCF_EXAMPLE's '##fileformat=VCFv4.3' line. Verified by extracting the PDF text locally.
3. 3
   Frequently Asked Questions: Data File Formats (BED, section format1; GTF, section format4)

   UCSC Genome Browser, Genomics Institute, University of California Santa Cruz

   TWO distinct things on the page. (a) BED: the entire BED_COLUMNS table, all 12 fields in the page's exact order and naming (chrom, chromStart, chromEnd, name, score, strand, thickStart, thickEnd, itemRgb, blockCount, blockSizes, blockStarts), the score range 'A score between 0 and 1000', the 'three required fields and nine additional optional fields' that backs the BED3/BED6/BED12 footnote, and the coordinate claims: "The first base in a chromosome is numbered 0" and "The chromEnd base is not included in the display of the feature. For example, the first 100 bases of chromosome 1 are defined as chrom=1, chromStart=0, chromEnd=100, and span the bases numbered 0-99" — which backs the COORDS BED row, the 'BED is 0-based and half-open' prose, and FAQ item 3's 'chr1 0 100 covers the first 100 bases' example verbatim. (b) GTF: the GTF_EXAMPLE and the GTF half of the GFF3/GTF section — that GTF shares the first eight fields with GFF, and the attribute syntax with its two mandatory attributes: "gene_id value - A globally unique identifier for the genomic source of the sequence" and "transcript_id value - A globally unique identifier for the predicted transcript", with "Attributes must end in a semi-colon, and be separated from any following attribute by exactly one space." Its GFF section also states "The first base is numbered 1" and end is "inclusive", backing the COORDS GFF3/GTF row.
4. 4
   Generic Feature Format Version 3 (GFF3) specification

   The Sequence Ontology

   The GFF_COLUMNS table using GFF3's own field names — seqid, source, type, start, end, score, strand, phase, attributes, in the page's exact order — and the GFF3_EXAMPLE. Backs column 8 'phase — Reading frame for CDS (0/1/2)': "The phase is one of the integers 0, 1, or 2, indicating the number of bases forward from the start of the current CDS feature the next codon begins." Backs the COORDS GFF3/GTF row and the '1-based and closed' prose: "The start and end coordinates of the feature are given in positive 1-based integer coordinates, relative to the landmark given in column one." Backs FAQ item 4's description of column 9 as tag=value pairs separated by semicolons, and the ID= / Parent= linkage shown in GFF3_EXAMPLE.

Related tools and references