Restriction Digest Calculator — Cut Sites and Fragment Sizes

Background.

This restriction digest calculator finds every occurrence of a recognition sequence in your DNA and reports the fragment sizes you would expect to see on a gel. It handles linear fragments and circular plasmids separately, because the answer genuinely differs: n cuts give n+1 fragments on a linear molecule but only n on a circle, and a site that straddles the arbitrary point where you started writing out a plasmid sequence is completely invisible to a linear scan.

One thing this page deliberately does not have is a dropdown of enzyme names, and the reason is worth stating plainly. REBASE, the curated database that is the authority for recognition sequences and cut positions, was consulted while building this page — version 607, dated 29 June 2026. Two automated reads of the same REBASE file, minutes apart, returned contradictory cut positions for the same enzymes: KpnI came back once as GGTAC^C and once as G^GTACC, SmaI once as CCC^GGG and once as C^CCGGG. Those are not typographical variations. They are different cuts of the same recognition sequence, and they produce different fragment sizes. Rather than ship a table of values that could not be verified, this calculator asks you for the recognition sequence and the cut position and tells you to take both from REBASE for your specific enzyme.

That is not merely defensive. It reflects a real hazard: the recognition sequence does not determine the cut. Enzymes that recognise the same sequence are isoschizomers, and those among them that cut it at a different position are neoschizomers — so two enzymes can share a six-base site and give you different band sizes. If you look up only the sequence and assume the cut, you can be wrong without any warning. Look up both.

The one default that does ship, EcoRI at G^AATTC, is verified from peer-reviewed primary literature rather than from those failed database reads. Dugaiczyk, Boyer and Goodman's 1975 paper in the Journal of Molecular Biology describes EcoRI fragments as terminated at their 5′ ends by the single-stranded sequence pA-A-T-T-, and a 5′ AATT overhang inside the palindrome GAATTC is only possible if the top-strand cut falls immediately after the G. The calculator's test suite asserts exactly that: the four bases immediately 3′ of every cut must read AATT.

What you get is a map, not a simulation. The calculator assumes every site is cut and cut completely. It does not model Dam, Dcm or CpG methylation, which routinely blocks a site from cutting at all and explains far more surprising gels than arithmetic ever does. It does not model star activity, partial digestion, buffer or temperature compatibility for double digests, or Type IIS enzymes such as BsaI that cut a fixed distance outside their recognition sequence. It models the top-strand cut only, so it reports fragment lengths — which is what a gel measures — and makes no claim about the overhang chemistry at the ends.

What is restriction digest calculator?

A Type II restriction endonuclease binds a short, usually palindromic, DNA sequence and cuts both strands at a fixed position relative to it. EcoRI recognises GAATTC and cuts between the G and the first A on each strand, leaving a four-base 5′ AATT overhang; a blunt cutter such as one recognising CCCGGG and cutting in the middle leaves no overhang at all. Because the recognition sites are short, they occur by chance at predictable frequencies — a six-base site appears roughly once every 4⁶, or about 4,100 base pairs, in a random sequence — which is what makes restriction mapping useful.

Digesting a molecule with a single enzyme produces a set of fragments whose sizes depend on where the sites are and on whether the molecule is linear or circular. On a linear template, n cuts produce n+1 fragments, because the two ends are already free. On a closed circle, n cuts produce exactly n fragments, because the first cut only opens the circle into a line rather than separating anything. A plasmid with a single site therefore runs as one band at full length after digestion, not as two.

Enzyme names follow the convention proposed by Smith and Nathans in 1973 and revised by Roberts and colleagues in 2003: an abbreviation of the source organism's genus and species, sometimes a strain designation, and a Roman numeral distinguishing multiple enzymes from the same organism. Two enzymes recognising the same sequence are isoschizomers; if they cut it at different positions they are neoschizomers. That distinction is why a recognition sequence on its own is not enough information to predict a digest, and why this calculator takes the cut position as a separate input.

How to use this calculator.

1. 01Paste your sequence. Plain text or FASTA both work — headers, whitespace, line numbering and gap characters are removed, case does not matter, and U is read as T.
2. 02Look up your enzyme in REBASE (version 607, 29 June 2026) and copy BOTH the recognition sequence and the cut position. Do not assume the cut from the sequence: neoschizomers share a site and cut it differently.
3. 03Enter the recognition sequence. You can paste it with the caret, as G^AATTC — the caret is stripped for matching, but you still need to set the cut position field yourself.
4. 04Set the cut position: how many bases of the recognition sequence lie 5′ of the top-strand cut. G^AATTC is 1, a blunt CCC^GGG is 3, a 3′-overhang CTGCA^G is 5, and a site cut before its first base is 0.
5. 05Choose the topology. Circular for a plasmid, linear for a PCR product or a gel-purified fragment. If you are unsure, try both — the difference in fragment count tells you a lot.
6. 06Read the fragment sizes, largest first. That is the band pattern you would expect. Check the sum: fragment sizes always add up to the template length exactly.
7. 07Sanity-check against a real gel. Bands within about 10 percent of each other will not resolve on standard agarose, and fragments under roughly 100 bp often run off the end or stain too faintly to see.
8. 08If a site does not cut in the tube, suspect methylation before suspecting the map — Dam, Dcm and CpG methylation block many enzymes and are not modelled here.

The formula.

The calculation is a scan followed by a set of differences. Every position where the recognition sequence matches is recorded — overlapping matches included, since a site such as AA genuinely occurs three times in AAAA. Each match at position s produces a cut after s + offset bases of the molecule, where offset is how many bases of the recognition sequence lie 5′ of the top-strand cut.

For a linear molecule the cut positions are sorted and the fragments are the gaps between them, plus the two ends: cut₁, then cut₂ − cut₁, and so on, finishing with L − cutₙ. That gives n+1 fragments from n cuts. Cuts landing exactly at position 0 or position L are discarded, because a cut at the very end of a linear molecule separates nothing — this is why the calculator never reports a zero-length fragment.

For a circular molecule the sequence is searched doubled, so a site straddling the origin is found, and cut positions are taken modulo the length. The fragments are the gaps between consecutive cuts, closing the loop with (L − cutₙ) + cut₁. That gives n fragments from n cuts. Zero cuts leave an intact circle and one cut merely linearises it, so both cases return a single fragment of the full length.

Ambiguity codes are handled conservatively. A symbol in the recognition sequence matches a template base when the template base's possible set is contained in the site symbol's set: a site R matches a template A or a template G, and a site N matches anything. The reverse does not hold — a template N is never matched by a concrete site symbol, because an unknown base is not evidence of a site.

Work through the loaded example. The 40 bp sequence AAAAGAATTCAAAAAAAAAAGAATTCAAAAAAAAAAAAAA has GAATTC starting at 0-based positions 4 and 20. With offset 1 the cuts fall after 5 and 21 bases. Treated as linear, the fragments are 5, then 21 − 5 = 16, then 40 − 21 = 19, so the gel shows three bands of 19, 16 and 5 bp, summing to 40. Treated as circular, the same two cuts give 21 − 5 = 16 and (40 − 21) + 5 = 24 — two bands, still summing to 40. Same sequence, same enzyme, different answer, which is why topology is a required input rather than an assumption.

Rounding: there is none. Every numeric output is an exact integer count of base pairs or of sites, and no floating-point arithmetic is performed anywhere in this calculator.

One limitation worth stating precisely, because it is easy to misread the output. Only the top-strand cut is modelled. Digesting the reverse complement of a sequence at the same offset is not the same operation — for a staggered enzyme it corresponds to the bottom-strand cut, which sits at (site length − offset) on the top strand, four bases away for EcoRI. For a blunt cutter, where the offset is exactly half the palindrome, the two coincide. This is why the page reports fragment lengths, which is what a gel measures, and says nothing about end chemistry.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Roberts, R. J., Vincze, T., Posfai, J. & Macelis, D. 'REBASE: a database for DNA restriction and modification: enzymes, genes and genomes.' Nucleic Acids Research 51(D1): D629–D630 (2023). doi:10.1093/nar/gkac975, PMID 36318248. Database served at rebase.neb.com; version 607, dated 29 June 2026, confirmed on retrieval 2026-07-29. Cited as the authority a reader must consult for their own enzyme's recognition sequence and cut position. Deliberately NOT used as the source of a built-in enzyme table on this page: two automated reads of the same REBASE file minutes apart returned contradictory cut positions (KpnI as GGTAC^C and as G^GTACC; SmaI as CCC^GGG and as C^CCGGG), and unverified values are not shipped.
2. [2]Dugaiczyk, A., Boyer, H. W. & Goodman, H. M. (1975). 'Ligation of EcoRI endonuclease-generated DNA fragments into linear and circular structures.' Journal of Molecular Biology 96(1): 171–184. The primary-literature verification of the one default this page ships: EcoRI fragments are described as 'terminated at their 5′-ends by the single-stranded sequence pA-A-T-T-', which fixes the top-strand cut immediately after the G in GAATTC. Asserted as a test — the four bases 3′ of every cut must read AATT. Paywalled; listed bibliographically.
3. [3]Roberts, R. J. et al. (2003). 'A nomenclature for restriction enzymes, DNA methyltransferases, homing endonucleases and their genes.' Nucleic Acids Research 31(7): 1805–1812. doi:10.1093/nar/gkg274. Source of the enzyme-naming convention and of the isoschizomer / neoschizomer definitions on which this page's central warning depends — that enzymes sharing a recognition sequence may cut it at different positions, so the sequence alone does not determine the fragment sizes. Open access via PubMed Central.
4. [4]Smith, H. O. & Nathans, D. (1973). 'A suggested nomenclature for bacterial host modification and restriction systems and their enzymes.' Journal of Molecular Biology 81(3): 419–423. Independent second authority consulted as a cross-check on the naming and recognition-site conventions used here, predating and independent of REBASE's curation; the Type II model of a recognition sequence plus a fixed cleavage position, written with a caret inside the site, is established there and unchanged in the 2003 revision. Paywalled; listed bibliographically.
5. [5]NC-IUB — Nomenclature Committee of the International Union of Biochemistry (1985). 'Nomenclature for incompletely specified bases in nucleic acid sequences: recommendations 1984.' European Journal of Biochemistry 150: 1–5, and simultaneously in Nucleic Acids Research 13: 3021–3030, Biochemical Journal 229: 281–286 and three further journals in 1986. The ambiguity alphabet accepted in both the template and the recognition sequence on this page. Read on the official IUBMB nomenclature site. Retrieved 2026-07-29.