Engine
CRISPR guide design
Choose a target gene; get back a ranked table of candidate guides you can sort, filter, shortlist and export. The scoring follows the approach established by CHOPCHOP and CRISPOR, and the table is the product — nothing is hidden behind a download.
1,000
catalog constructs
65
host organisms
CPU
engine class
Free
to run
01
What the feature does
Guide selection is a bounded problem with a tedious middle. The candidates for a given target are enumerable and the criteria for ranking them are well established; what actually consumes the afternoon is getting tooling installed, pointed at the right genome build, and producing a result anyone can read. The guide designer removes that middle and leaves you with the judgement call.
You select a gene. The engine enumerates every candidate guide in the window you chose and returns them as a ranked table — one row per candidate, with an on-target score, GC content, position relative to the coding sequence, and strand. Ranking is visible rather than implied, which matters, because the top row is a suggestion and you are allowed to disagree with it.
02
In the interface
Pick a target from the catalog
Targets come from the 1,000-construct catalog or from your own project. Entering a catalog ID such as SS-0421 loads the construct with its host and its annotated features already attached, so the region you are aiming at is described before you choose anything.
Set the enzyme and the window
A dropdown covers the common nuclease variants and their PAM requirements. The search window can be a whole coding sequence, a named feature from the annotation, or a coordinate range you type in — the engine enumerates candidates on both strands inside it.
Read the ranked table
Results arrive as a table ranked by on-target score, with GC content, position and strand in their own columns. Scoring follows the approach published for CHOPCHOP and CRISPOR, so a reviewer who knows those tools knows how to read this output.
Sort, filter, shortlist
Column sorts and threshold filters apply instantly and change nothing underneath: cap GC content, exclude candidates in the final stretch of the coding sequence, keep the top twenty by score. Filters compose, and the current view is exactly what exports.
Inspect a candidate
Any row expands into a detail panel showing where the candidate sits on the annotated map, which features it overlaps, and how its score was composed. Star the ones worth keeping; the shortlist travels with the project rather than living in a spreadsheet.
Export or pass it along
A shortlist exports as GenBank, FASTA, SBOL, or an IDT or Twist order sheet. It also passes straight into off-target analysis without re-entry, which is where most shortlists go next.
03
A worked example
- Target
- Catalog construct SS-0421
- Host
- Selected from the 65 supported organisms
- Search window
- Coding sequence, both strands
- Candidates enumerated
- 62
- After GC and position filters
- 18
- Shortlisted
- 6
- Runtime
- Under a minute
- Engine class
- CPU — no credits
Sixty-two candidates is an ordinary return for a target of this size, and the number is not the useful part. Two filters — a GC ceiling and a rule excluding the last stretch of the coding sequence — cut the table to eighteen rows in a couple of clicks, and the six that got starred were chosen by a person looking at the detail panel, not by the sort order.
That shortlist is the unit of work downstream. It carries into the specificity sweep, it appears in the construct record, and it shows up in the validation report attached to the design later. The table you exported and the guides that ended up in the project are the same six rows.
04
Who runs it, and what it costs
Breeding groups use it as the first pass on a new target, usually before anyone has committed to a host. Ag-biotech teams run it across a whole series at once through the batch queue and compare the score distributions rather than individual guides. Teaching labs use it because the output is legible — a student can see why one candidate outranks another without reading a manual first.
Guide design is a CPU engine, so it is free to run, with no per-run charge and no limit on exports. Prepaid credits apply only to GPU work elsewhere on the platform — $25 for 250 credits, one credit for roughly one Standard job — and nothing on this page draws on them.
Ready to run it
Try this in the Splice Platform.
CPU engines are free to run. GPU engines run on prepaid credits.
Related