Genome Coordinate System Converter — BED (0-based) ↔ GFF/VCF/SAM (1-based)

BED coordinates and GFF/VCF coordinates differ by 1 — this is the most classic source of off-by-one errors in bioinformatics. This tool converts an interval between the two coordinate systems and gives you ready-to-paste BED and GFF lines.

The Two Coordinate Systems

Both systems are explicitly defined in the specification (see sources at the end):

The canonical example from the spec itself: bases 3 through 7, inclusive, written as [3, 7] in 1-based notation, and [2, 7) in 0-based notation.

Only the Start Differs by 1 — the End Is the Same

This is the single most important thing to remember. Using the example above:

GFF    start = 3    end = 7
BED    start = 2    end = 7
                          ↑ identical

BED’s end and GFF’s end are numerically identical. The only difference is in start. The reason: the 0-based half-open interval shifts the start one position to the left, while the end is exclusive — the two offsets cancel out exactly.

The most common mistake is adding 1 to end as well, which silently extends every interval by one base. This produces no error message — it just skews your statistics quietly. One extra base per interval, tens of thousands of intervals, and the cumulative error becomes substantial.

The Length Formula Differs Accordingly

BED   length = end − start        (no +1)
GFF   length = end − start + 1    (add 1)

Both formulas give the same result. The example above is 5 bp either way: 7 − 2 = 5, 7 − 3 + 1 = 5. The classic symptom of mixing them up is that all interval lengths are off by 1 — if your calculated exon lengths are consistently 1 less or 1 more than database values, you have almost certainly applied the wrong formula.

Single-Base Sites

A SNP at VCF POS = 1000 becomes, in BED:

chr1    999    1000

It looks like it spans two positions, but the length is exactly 1000 − 999 = 1 — one base. When converting a SNP list to BED, subtract 1 from start; leave end unchanged.

BED Can Represent Zero-Length Intervals; GFF Cannot

In BED, start == end (e.g., chr1 100 100) is valid and represents a zero-length insertion point — an insertion occurring between base 100 and base 101.

A 1-based closed interval cannot represent zero length: [101, 100] (start > end) is illegal in GFF (the spec explicitly requires start ≤ end). Such intervals cannot be directly converted; the insertion site can only be described in text. This tool will tell you explicitly when a conversion is not possible, rather than returning a plausible-looking but incorrect result.

A Further Pitfall: Strand Orientation

Coordinate systems are just the first layer. BED column 6 and GFF column 7 both record strand, but coordinates are always given relative to the forward (plus) strand — a gene on the minus strand still has start < end; the two values are not swapped. When you need the “5′ end of the gene itself,” you must use strand to decide whether to take start or end.

Data Sources

The definitions of both coordinate systems and the format assignments come from the terminology section of the SAM specification in the official htslib spec repository (Samtools/hts-specs, SAMv1 §1), which explicitly states that “SAM, VCF, GFF and Wiggle use 1-based coordinates” and “BAM, BCFv2, BED and PSL use 0-based coordinates,” and gives the [3,7] / [2,7) example.

Two independent corroborations: the VCF specification (VCFv4.3) states “POS is the reference position, with the first base having position 1”; the GFF3 specification (Sequence Ontology) states coordinates are “positive 1-based integer coordinates” and “start must be less than or equal to end.”

Related Tools

For sequence composition statistics, see Basic Sequence Statistics; for sequencing depth and data volume, see Sequencing Coverage Calculator.