Free Sequence Tools for an iGEM Season: A Phase-by-Phase Checklist

What hasn't changed about the competition (and one thing that has)

The structural facts worth planning around are stable: iGEM runs as a team competition across three divisions — High School, Undergraduate, and Overgraduate. The Overgraduate cutoff is age, not degree status: a team competes in Overgraduate if any student team member is older than 23 at the registration deadline, and graduate students or postdocs associated with a team more often serve as advisors than as counted student members. Beyond that, teams work from a distribution kit of standard parts plus whatever they design themselves, new parts get submitted to the Registry of Standard Biological Parts, and every team documents its project on a public wiki that is itself judged — Best Wiki is a standing special prize, and medal and special-prize criteria reward showing your engineering reasoning, not just presenting a construct that worked. The Grand Jamboree is held in Paris in the fall.

One thing that has changed and is worth knowing before you bookmark anything: the classic parts registry at parts.igem.org is now in read-only archive mode. It's still fully browsable and searchable, but logins, edits, and new part submissions are disabled there. The actively maintained registry — where current Part Pages and this season's submissions live — is at registry.igem.org. If your wiki template or a previous team's page links to parts.igem.org for submission instructions, check the current registry site instead; a surprising amount of iGEM tooling advice online still points at the legacy URL.

The five checkpoints, and what covers them

Each of these has adequate free tooling. The rest of this guide goes through them in the order a season actually hits them.

• Primer design for PCR screening and for Sanger confirmation reads — melting temperature estimation, specificity, and read-spacing planning.
• Golden Gate/MoClo overhang design, checked for cross-reactivity before fragments are ordered — a fidelity score, not just a duplicate check.
• Construct linting before a synthesis order goes out — premature stops in the real post-assembly frame, cryptic motifs, and restriction sites that conflict with your assembly standard.
• Batch screening of the 15–20 parts a team typically pulls from the kit and registry each season, instead of running each one through a single-sequence web form.
• Verifying what actually grew back: aligning Sanger reads from returned colonies against the intended sequence, not just eyeballing a chromatogram.

Primer design: PCR screening and sequencing reads

Two different primer jobs come up in almost every iGEM project and get conflated more often than they should. Screening/cloning primers need to amplify one specific product cleanly; sequencing primers need to sit far enough back from a junction that the noisy first bases of a Sanger read don't land on the region you actually care about.

For melting temperature, the quick-and-dirty Wallace rule (Tm = 4°C × (G+C) + 2°C × (A+T)) is still useful as a sanity check you can do in your head, though nearest-neighbor methods (which use the stacking energy of each adjacent base pair rather than a flat per-base value) are more accurate and are what most modern primer-design tools use by default — the two methods can disagree by several degrees on the same primer, more so as GC content or length moves away from a 'typical' 18–22mer.

Worked example: take the 20-nt primer 5′-CAGCTGGACGGCGACGTAAA-3′. Counting bases: 12 are G or C, 8 are A or T, giving 60% GC. The Wallace estimate is 4×12 + 2×8 = 64°C. That's on the high end for standard PCR conditions and is exactly the kind of primer where you'd want its partner primer within a few degrees of the same estimate before ordering a pair, and where a nearest-neighbor recalculation is worth doing before you lock in an annealing temperature.

Beyond Tm, a primer-design pass before ordering should check: a GC clamp at the 3′ end, absence of a stable self-dimer or hairpin at your intended annealing temperature, and specificity against the whole construct (not just the local region) — a primer that also anneals somewhere else in your backbone or genome will give you a confusing secondary band on a screening gel that costs a day to debug. For sequencing primers specifically, plan for roughly 700–900 bases of usable high-quality Sanger read per primer; anything longer than that needs a second primer (internal or from the other direction) to get full coverage, and the first 20–50 bases immediately after a primer are typically the noisiest part of the read, so don't place a primer so close to your region of interest that the noisy window overlaps it.

Golden Gate and MoClo overhang fidelity: check cross-reactivity before you order fragments

Golden Gate and MoClo-style assembly work by cutting fragments with a Type IIS enzyme (commonly BsaI, recognition site GGTCTC, or BsmBI/Esp3I, recognition site CGTCTC) that leaves a defined 4-nt overhang at each junction, then relying on those overhangs to anneal correctly during ligation. The failure mode that's invisible until you've already paid for synthesis is cross-reactivity: an overhang that's identical, or close enough, to another overhang's reverse complement in the same assembly can pair with the wrong fragment, or in the wrong orientation.

This is a real, measured phenomenon, not a theoretical concern. Potapov et al. (Comprehensive Profiling of Four Base Overhang Ligation Fidelity by T4 DNA Ligase and Application to DNA Assembly, ACS Synthetic Biology, 2018) experimentally profiled T4 DNA ligase's fidelity across all 256 possible four-base overhangs and used that data to predict junction sets for higher-order Golden Gate assemblies. That dataset is the basis for most modern fidelity-scoring tools, including NEB's free NEBridge Ligase Fidelity Viewer, because it captures something a simple 'are these overhangs unique' check misses: some non-identical overhang pairs still mis-ligate at a meaningfully higher rate than others, for reasons that come down to real ligase biochemistry, not string matching.

Worked example of the string-matching part you can do by hand before reaching for a tool: suppose you've sketched four junction overhangs for a promoter–RBS–CDS–terminator assembly: AATG, CGCT, AGCG, and TTCG. Take the reverse complement of CGCT — complement each base (C→G, G→C, C→G, T→A) to get GCGA, then reverse it to get AGCG. That's exactly your third overhang. CGCT and AGCG are reverse complements of each other, meaning the ligase can join those two junctions in either 'orientation' interchangeably — a real design conflict, not a typo. Separately, a 4-nt overhang like GATC is its own reverse complement (self-palindromic: complement it to CTAG, reverse that back to GATC), which Golden Gate designs generally avoid because a palindromic overhang can anneal to itself in either direction. Catching AATG and TTCG required no special check here — they're distinct from each other and from every reverse complement in the set.

Doing this by hand does not scale past a handful of junctions, which is the actual argument for a fidelity-scoring tool rather than a spreadsheet: it needs to check every overhang against every other overhang's reverse complement, weighted by the real Potapov ligation-fidelity data, not just flag exact duplicates. This is exactly what SeqBench's Golden Gate overhang fidelity tool automates — score a full junction set programmatically through its REST API or MCP server so it fits into a design-iterate loop, or check a smaller set interactively in its general assembly simulator UI. It is not the only free option: NEB's NEBridge Ligase Fidelity Viewer, built on the same underlying dataset, is a solid alternative if you want a second source before committing to an order.

One iGEM-specific wrinkle worth knowing: the registry's current submission standards are the classic BioBrick RFC10 restriction/ligation standard and the newer Type IIS standard (RFC1000). RFC1000 is itself built on MoClo and Loop assembly conventions, so Golden Gate-family assembly does have official standing in the registry — but that doesn't mean any overhang set you design for your own Golden Gate build is automatically submission-ready. RFC1000 defines its own specific fusion-site sequences for basic parts, transcriptional units, and multi-transcriptional-unit levels, and it specifically treats internal BsaI and SapI sites as illegal within a submitted part, because those enzymes are reserved for the standard's own assembly junctions. If you're assembling with Golden Gate internally but plan to submit a part, check which submission standard applies (RFC10 or RFC1000) and lint for the right set of forbidden sites — see the next section.

Linting a construct before it goes to a synthesis vendor

A construct that looks fine as a set of individual parts can pick up problems only after assembly: a reading frame that shifts by one base at a Golden Gate scar, a cryptic ribosome-binding-site-like sequence created at a junction, or a restriction site that's fine in isolation but now conflicts with the assembly standard you're submitting under. Linting before submission means checking the final, assembled sequence — not each input fragment separately — for a specific set of problems:

• Premature stop codons in the actual reading frame the assembled construct will use, checked on both strands if there's any chance of a cryptic reverse-frame ORF overlapping a promoter or regulatory region.
• Restriction sites relevant to your chosen standard: EcoRI, XbaI, SpeI, PstI, and NotI for classic BioBrick RFC10 (the prefix/suffix enzymes), or BsaI and SapI for the Type IIS RFC1000 standard.
• Cryptic motifs that can cause unintended expression or termination: internal Shine-Dalgarno-like sequences that could initiate translation from the wrong start codon, rho-independent terminator-like hairpins sitting in the middle of a coding sequence, and unintended promoter-like elements.
• Vendor-level synthesis constraints that cause rejected or delayed orders regardless of biology: extreme GC content (very high or very low stretches), homopolymer runs of five or more identical bases, and strong hairpin secondary structure. These are worth checking even for a construct that passes every biological check, because a rejected synthesis order is a real timeline cost — typically another design-review-and-resubmit cycle — that an unfunded team on a fixed calendar can't easily absorb.

Batch-processing many parts instead of one at a time

A typical iGEM project touches somewhere in the range of 15–20 sequences in a season — parts pulled from the distribution kit, parts pulled from the registry, and a handful of new designs — and most of those need the same handful of checks: GC content, Tm for their flanking primers, and a restriction-site/motif scan. Running each one individually through a single-sequence web calculator means repeating the same few clicks 15–20 times, which adds up to real, avoidable hours over a season that's already short on them.

The fix is running the same check across all of them at once instead of one at a time. SeqBench's batch-processing endpoints accept a FASTA of many sequences and apply the same tool — Tm calculation, GC content, restriction-site scan, or others — across the whole set in a single call, which is the kind of thing worth citing specifically on a wiki methods page: naming the batch tool and what it was run against is more useful documentation than describing the check in the abstract, because it tells a judge (or another team trying to reproduce your work) exactly what was and wasn't verified.

Verifying returned colonies against the intended sequence

Once fragments come back from synthesis or a cloning reaction produces colonies, the step that's easiest to shortcut under deadline pressure is also the one that catches the most expensive mistakes: actually comparing the Sanger read to what you intended to build, rather than confirming a colony grew and moving on.

1. Design sequencing primers with enough buffer (roughly 150–250 bases) before the region you actually need to confirm, since the bases immediately following a primer are the noisiest part of a Sanger read.
2. For inserts longer than about 700–900 bases, use more than one sequencing primer (walking or bidirectional) so every base of the region of interest is covered by at least one high-quality read.
3. Trim each raw read for quality before aligning it — don't align the full untrimmed read, since low-quality tails will generate false mismatches.
4. Align each trimmed read against the full intended construct, not just the amplicon you expect to have sequenced. This is what catches a wrong insert, backbone-only self-ligation, or an unexpectedly duplicated fragment — problems a narrow, region-only comparison can miss entirely.
5. If the construct was Golden Gate-assembled, specifically check the junction/scar sequence at each overhang position against the intended fusion site. A single wrong base at a junction can still transform and grow into a colony; it just won't behave as intended.
6. Treat mismatches inside homopolymer runs of four or more identical bases, and mismatches within the first ~20–50 bases after the primer, with extra scrutiny before calling them real — both are known weak points of Sanger base-calling, not necessarily real sequence variants.
7. Record which colonies passed and which failed, against which construct version, and note the alignment tool and parameters used — this is exactly the kind of methodological detail that counts as documented engineering work on a wiki, rather than a bare 'confirmed by sequencing' claim.

What none of these checks tell you

Before any of the above goes into a wiki as 'verified,' it's worth being explicit about what each check does not cover:

• A high Golden Gate fidelity score is a prediction from in vitro data generated with purified fragments and a specific ligase under defined conditions (the Potapov et al. dataset). It doesn't account for your actual insert concentrations, incomplete digestion, star activity of the Type IIS enzyme at near-cognate sites, or a bad enzyme lot. Even a strong score still means screening real transformants, not skipping that step.
• Restriction-site and motif linting only flags what it was told to look for. It won't catch every possible cryptic regulatory element a full genome-annotation pipeline might flag — it's a targeted sanity check against a known list (specific enzymes, common terminator/RBS-like patterns), not a functional annotation of the construct.
• A premature-stop-codon scan depends entirely on you specifying the correct reading frame and strand for the assembled construct. A frame that's correct in the plasmid map you assumed but shifts by one base because of an actual Golden Gate scar will pass a scan run on the wrong frame and still fail at the bench.
• Tm predictions, whether Wallace-rule or nearest-neighbor, assume idealized buffer and salt conditions and no secondary structure. Neither method will tell you that a specific primer forms a hairpin at your actual annealing temperature on your actual thermocycler — a temperature gradient on a new, important primer pair is still worth the plate.
• Comparing a Sanger read to your intended sequence confirms that read's covered region, from one colony, on the strand you sequenced. It does not confirm plasmid copy number, rule out a mixed colony population, or catch multimerized/recombined inserts that a single amplicon-level check wouldn't reach — that needs full coverage across the construct, not a spot check.
• None of this substitutes for growing out and functionally testing the construct. A construct with a clean linting pass and a strong overhang fidelity score can still fail in ways no sequence tool checks: toxicity of the expressed part, an RBS strength that's wrong for your chassis, or simply too little of the plasmid to detect.

Making tool use count on the wiki

iGEM judging rewards documented reasoning, and a wiki methods section that names the actual tool, version, parameters, and result is worth more to a judge than a sentence like 'primers were designed using standard methods.' A short, specific paragraph — which tool, what it checked, what threshold or parameter you used, and what it found, including negative results like 'no internal BsaI sites were found in any of the 18 parts screened' — is both faster to write than it sounds and directly answers the kind of question a judge or a future team re-using your wiki will actually ask.

Related tools

Related guides

• Gibson Assembly vs. Golden Gate vs. Restriction Cloning: Choosing a Strategy →
• Cloning a Gene From Scratch: A Complete Workflow From Sequence to Verified Construct →
• Golden Gate Overhang Fidelity, Scored Against Real Ligation Data: Three Published Sets vs. Two Naive Designs →