How to Open an .ab1 File (Without Installing Anything)

An .ab1 is four traces, not a sequence

This is the thing that makes the format confusing. A FASTA file contains a sequence. An .ab1 contains the raw output of a capillary sequencer: four channels of fluorescence intensity sampled across the length of the run, one per dye. The base calls are in there too, but they are a derived annotation on top of the traces, and so is the per-base quality.

The container is called ABIF, which the specification's own historical notes expand as "Applied Biosystems, Inc. Format". Its specification was published by Applied Biosystems — the latest revision is titled "Applied Biosystems Genetic Analysis Data File Format", dated September 2009 — and it opens with an unusually blunt disclaimer: "Applied Biosystems does not support users of this specification in any way. Please do not call technical support for additional information pertaining to this specification." The document is also no longer served from its original vendor URL, which now redirects to Thermo Fisher's Applied Biosystems brand page.

Structurally it is a generic tagged container, closer to TIFF than to a sequence format. The first four bytes are the ASCII characters A, B, I, F, which is how a reader confirms the format. Every integer is big-endian. A fixed 128-byte header at the front points at a directory of 28-byte entries, and each entry names a tagged data item — the base calls, the quality values, each trace channel, the sample name, the instrument settings, the run start time, and dozens more.

One detail matters more than it looks: ABIF defines no fixed vocabulary of tags. Which tags a file carries depends on the instrument and the software version that wrote it, and the spec only notes that the container is designed for fewer than a thousand data items. Real instrument files from the 310, 3100 and 3730 generations carry roughly 107 to 130 directory entries, while stripped-down test fixtures can carry as few as 19. The portability problem does not rest on that spread, though — it rests on the spec saying outright that the format "does not specify the schema ... which tags are written and when", which is why a reader written against one lab’s files can fail on another’s.

.ab1 and .abi are the same format. .scf is not

People are routinely told these are three variants of one thing, and get sent in circles as a result. Two of them are one thing and the third is unrelated.

.ab1 and .abi are the same ABIF format under two extensions. The vendor's own specification is the proof, and it is slightly embarrassing about it: within a single document, the same tag table is introduced as "ab1 File Tags" in its section’s own list of tables and captioned "abi File Tags" on the page itself. The earlier 2006 revision labels the analysis tag tables "abi File Tags" where the 2009 revision calls the same material tags added to ".ab1" files. If a tool reads one, it reads the other. There is also .fsa, which is the same ABIF container written by fragment-analysis runs rather than sequencing runs.

SCF is a genuinely different format — the Standard Chromatogram Format, from the Staden package. It has its own specification, its own magic number (the ASCII characters ".scf"), and its own 128-byte header layout. Its lineage runs the opposite way to what you might guess: Dear and Staden described it in 1992 precisely because vendor formats already existed and were mutually incompatible — their paper names the Applied Biosystems 373A and the Pharmacia A.L.F. — so SCF is the later, vendor-neutral standardisation attempt rather than something ABIF displaced. It happens to also be big-endian, which is the only thing the two have in common.

So "just rename it to .scf" cannot work, and neither can the reverse. Converting between them needs a tool that understands both, and even then the traffic is one-way in the free tooling: Staden's own io_lib ships a convert_trace utility that reads ABIF and writes SCF, but its ABIF write path is commented out in the source, so it cannot go back the other way.

Why renaming or copy-pasting doesn't work

Renaming pUC19-fwd.ab1 to .txt, .fasta or .seq changes nothing about the bytes. An extension is a hint to your operating system about which application to launch; a FASTA parser fed an ABIF file still looks for a line starting with ">", hits binary data, and fails.

Opening it in a text editor is more tempting, because it half works. ABIF stores the base calls as plain ASCII letters, so among the noise you will often see a long readable run of ACGT, plus a recognisable sample name. Copying that out is not a real answer: there is no reliable way to tell by eye where the base calls start and stop, you will silently lose the per-base quality that tells you which of those calls to trust, and you cannot tell whether you grabbed the primary base calls or a second copy of them that analysis software also wrote into the file.

There is one thing a text editor is genuinely good for here, which is confirming the file is not corrupt. If the first four characters are ABIF, the file is structurally an ABIF file and your problem is a reader, not a bad download.

Open it in a browser

SeqBench's Sanger Trace Viewer opens .ab1 and .abi in a browser tab with no install and no account — the file is uploaded, parsed, and never logged or stored. It decodes the ABIF directory, maps the four DATA channels to bases using the file's own FWO_ (filter wheel order) tag rather than assuming an order, and draws the electropherogram with the called bases positioned over their peaks. It reports read length and mean Phred quality, and exports the base calls as FASTA or the trace figure as SVG or PNG.

Being straightforward about the limits, because they are the difference between this being the right tool and the wrong one. It reads ABIF only — not SCF. It exports FASTA but not FASTQ, and it summarises the per-base quality it parses as a single mean rather than showing it base by base — so if you need the quality values themselves, use one of the FASTQ routes below. It does no quality trimming. It takes one file at a time, so it is not the answer for a folder of ninety-six reads. And it does not flag heterozygous or mixed positions for you — it draws the peaks and you read them.

If what you need is any of those things, one of the options below is a better fit, and it is worth saying so plainly rather than sending you round a loop.

The other free browser options, and what each is actually for

One thing to be aware of while searching. At least one site currently ranking for "ab1 viewer" is a re-skin of GEAR Teal that relays your upload to a third-party Cloudflare Worker — its responses to the same file are byte-for-byte identical to EMBL's. It works, but you are handing your data to an unrelated intermediary for a service EMBL provides directly. Prefer the original.

• EMBL's GEAR Teal (gear-genomics.com/teal) — a trace viewer that returns both a primary and a secondary base sequence, per-base quality and all four peak channels. Free, no login, run by EMBL. Two caveats: it exports by copy-to-clipboard only, with no FASTA download, and their terms say submitted data is held on EMBL's server for up to 24 hours before deletion. Link the www host; the bare domain does not respond.
• EMBL's GEAR Indigo — the right tool if your question is about a mixed or heterozygous trace. It decomposes a trace into two alleles against a reference, calls and annotates the variants, and estimates allele fractions. Note that it requires a reference (a genome from its dropdown, a FASTA, or a wild-type trace), so it is not a drop-in-one-file viewer, and its outputs are a PDF report and VCF/BCF rather than FASTA.
• chiplot.online's ab1 viewer — worth knowing about, but not currently usable: the page still loads, yet its upload endpoint returned 502 on every attempt when checked in August 2026, so it does nothing with your file. Re-test before relying on it.
• bugaco's sequence converter — not a viewer at all, but it does convert ABIF straight to FASTA or FASTQ with no account, which makes it the quickest no-code route to a FASTQ that keeps your quality scores. Two traps worth knowing: GenBank sits in its output dropdown but throws a Python error on ABIF input (EMBL output works), and its "abi-trim" option can return a FASTA header with no bases at all, so check the output is non-empty before trusting it. It is a personal side project rather than an institutional service.
• TIDE, from the van Steensel lab at the Netherlands Cancer Institute — narrow but excellent at its job: give it a control trace, a test trace and a guide sequence and it quantifies a CRISPR indel spectrum from the chromatograms.

The desktop tools you'll be told to use, and their real state

Most advice on this question is inherited from lab notes written years ago. Here is where those recommendations actually stand, checked against vendor sources rather than download aggregators.

• FinchTV — the most-recommended and the most dead. It appears in its vendor's own product lifecycle document in the End of Life table, all versions, with an end-of-life date of December 2016; that document says end-of-life versions are "no longer available to the market", and the Sunset stage before it is already defined as no longer sold. Geospiza's original site, which hosted it, no longer answers HTTP at all. The last release was 1.5.0 and only for Intel Macs — Windows stopped at 1.4.0, Linux and Solaris at 1.3.1. It was free but gated behind a form asking for your name, email, phone and organisation. For the record it did read both formats: Geospiza's archived help pages list "View ABI and SCF chromatogram files" and "Save traces in ABI or SCF format", though those pages are now reachable only through the Wayback Machine. It may still run; nobody supports it, and no vendor download remains.
• 4Peaks — Mac only, and dormant rather than dead. The vendor lists version 1.8 requiring "OS X 10.7 and above", and states it reads "formats like AB1 and SCF" including quality data. But the download binary's last-modified date is 13 July 2015, the site's copyright line stops at 2017, and the vendor says nothing anywhere about 64-bit or Apple Silicon. It is labelled free only by a badge generated in the site's stylesheet; there is no pricing page, licence page or EULA at all.
• Chromas — genuinely free, in the vendor's own words ("Chromas is free of charge"), currently 2.6.6, and the most capable of these for conversion: it opens .ab1, SCF and ZTR, saves as .scf or .ab1, and exports sequence as plain text, FASTA, FASTQ, EMBL, GenBank or GCG. The catch is that it is Windows only. Two things not to be confused by: the optional PeakTrace RP component inside it becomes a paid service after 40 free units, and the separate ChromasPro is paid software (USD 16/month or 299 perpetual, with academic rates). "Chromas Lite" no longer exists as its own product — that URL now redirects to the free Chromas page.
• Thermo Fisher's Sequence Scanner Software 2 — still listed by Thermo in 2026 as a Sanger "Viewer" and described as freeware, so it is free. It is Windows only, and the only OS Thermo claims for it is Windows 7, which is a meaningful signal given that neighbouring entries in the same table list Windows 10 and 11. The download also sits behind a mandatory registration form asking for your name, job role, telephone, email, institution, city and country. Notably, Thermo's page never actually says which formats it reads.
• SnapGene — reads trace files in its free Viewer mode. SnapGene and the old separate Viewer are now one application: viewing, annotating and sharing are free, while editing, aligning and cloning need a licence. The free download requires an email address and agreeing to the Terms of Use.

If you're already scripting

Biopython is the standard route, and the details are worth getting right because two of them surprise people.

The format strings are "abi" and "abi-trim" — not "abif", which appears in Biopython's own older release notes but is not a valid argument and will raise an unknown-format error. Reading has been supported since Biopython 1.58; "abi-trim" arrived in 1.71. Both are read-only: there is no ABI writer, and attempting one raises ValueError with "Reading format 'abi' is supported, but not writing".

The useful part is that the reader does expose per-base quality, pulled from the ABIF quality tag into the record's phred_quality annotation — which means you can convert an .ab1 directly to FASTQ and keep the quality scores, something most of the browser viewers will not do for you. Every raw ABIF tag is also stashed on the record, including the DATA channels, so you can plot the traces yourself without another library. One gotcha for file naming: the record id comes from the file's internal sample-name tag, not from the filename, so a FASTQ written from a folder of traces is headed with sample names.

The "abi-trim" variant implements Richard Mott's modified trimming algorithm with hard-coded constants — a minimum length of 20 and a cutoff of 0.05, giving a per-base score of 0.05 minus that base's error probability, which turns positive at about Q13. Records of 20 bases or fewer come back untrimmed, and a file carrying no quality values raises an error rather than silently passing through. Expect it to be aggressive: on one real heterozygous trace, trimming cut a 605-base record to 183 bases.

In R, sangerseqR is current and maintained, and unlike most of the options here it does secondary-peak calling — makeBaseCalls identifies windows containing more than one peak and makes heterozygous calls, filling in both a primary and a secondary sequence. For whole sets of reads, sangeranalyseR builds on it with trimming, indel detection and reports.

Converting a whole folder

This is a common need that almost none of the browser tools serve, so it is worth naming the honest answers. All of these require installing something; there is no good no-install batch route.

The shortest single command is Staden io_lib's extract_fastq, which accepts multiple filenames and concatenates every record into one output, so extract_fastq *.ab1 > all.fastq genuinely works. It is free, BSD-style licensed, and still maintained. Note it names each record after the input file's basename rather than the internal sample name — the opposite of Biopython's behaviour.

The most controllable route is a short Biopython loop over a glob calling SeqIO.convert, which gives you FASTA or FASTQ per file or combined. If you want better basecalling and trimming than a plain conversion, tracy (the engine behind GEAR Indigo, and actively released) can emit FASTA or FASTQ per trace, though it takes one file per invocation so you still need a shell loop.

One tool to be careful about: DNA Baser markets an "ABI to FASTA Converter" whose page title calls it free, but its own licence section says the free version converts only a limited number of files and does not say how many. The tool is listed at USD 44.90, the assembler starts around USD 69, and all of it is Windows only. EMBOSS can read ABIF too, but its newest release tarball is dated July 2016, so treat it as frozen.

What to check once it opens

Two sanity checks before you read anything into the sequence.

First, is the read as long as it should be? On a modern capillary instrument a Sanger reaction typically returns on the order of 700 to 900 bases of usable, high-quality sequence. Applied Biosystems rates the 3730xl at 850 QV20 bases on its two-hour long-read module and 900 on the three-hour extra-long module, and core facilities running 3730-class instruments publish figures anywhere from about 600 to 900 depending on template, chemistry and run module. A read that comes back at 200 usable bases points at a problem with the template or the primer, not with your file reader.

Worth knowing what that published number means, because it explains something that otherwise looks like a bug: the specified figure is a continuous read length — the longest uninterrupted stretch whose average quality value stays at 20 or better, computed over a sliding window. Your file will normally contain called bases past the end of that stretch. Those trailing calls are real output, they are just not covered by the instrument's quality specification.

Second, ignore the start of the read. The first 20 to 35 bases are the least trustworthy part of any trace, and it is a resolution limit rather than a failed reaction. Applied Biosystems puts it directly: the properties of the dyes and the separation polymer together limit the ability to resolve the first 20 to 25 bases, and poor quality through the first 35 is normal with its standard chemistry and polymer. Its own suggested remedy is a chemistry swap, which is the tell that this is not something a rerun of the same reaction fixes. Quality falls away again at the far end as peaks broaden and merge, and the 3500-series specification makes the gap concrete: for its standard sequencing module the resolution range committed to in at least 90% of samples starts by base 40 and extends to at least base 700, against at least 850 QV20 bases for that same module.

A related artefact people misattribute: dye blobs, the broad off-colour humps from unincorporated dye terminators, sit at roughly 85 to 90 bases with standard chemistry, and at roughly 60 to 65 and 125 to 140 bases in more severe cases. They are downstream of the noisy opening window, not the cause of it, and their presence points at a cleanup problem rather than something inevitable.

Once the file is open and you are deciding which specific peaks to believe — whether a stacked double peak means a mixed clone, whether a mismatch is real — that is the subject of the clone-verification guide, and the Sanger vs Reference tool will align your read against its expected sequence and list the differences for you.

How to not hit this again

Ask your sequencing core whether they can also deliver FASTA or FASTQ alongside the traces. Most can, many do by default, and it removes the problem for every routine check where you only need the sequence.

Keep the .ab1 files anyway. The trace is the evidence, and it is the only thing that can settle an ambiguous call later — a FASTA of the base calls silently discards the four channels of raw signal that let you see a double peak, a broadened region or a dye blob for what it is. Archive the traces and work from the extracted sequence.

Related references

Related tools

Related guides

• How to Verify a Clone with Sanger Sequencing →
• How to Calculate GC Content of DNA (Formula) →
• Reverse Complement of DNA, Explained →