CRISPR guides

The CRISPR tab finds guide RNA target sites: every protospacer next to a PAM, on both strands, including sites that cross the origin of a circular sequence. The scan runs in the background and follows the document, so an edit gives a new list without asking.

Each row shows the strand (→ forward, ← reverse), the spacer with its PAM run on in grey, the GC content, and the off-target counts. A row with a ! has something worth a look — hover it for what. Clicking a row selects the protospacer in the views and opens the guide's details right under the row, so the list stays in view; clicking another row moves them there, and clicking the open row again closes it and clears the selection.

While the tab is open, every guide listed is drawn on the map and in the sequence view as an arrow on its own strand labelled with its PAM (see Previews); clicking one there opens it here and scrolls the list to it. Past 200 guides none are drawn and only the first 200 are listed — filter the list or narrow the scan to a selection to see the rest.

Filtering the list

The tab is two groups: Options sets what is scanned for, and Guides lists what was found. The filters at the top of Guides work on the guides already found, so they are instant and the group's title says how many are shown, as in 139 of 345:

• Spacer contains… keeps the guides whose spacer has the bases typed, in IUPAC codes as in Find — GRCC or TTTN work — which is how to find a guide from a paper, or one starting with a G for a U6 promoter.
• PAM keeps one PAM out of those found: SpCas9's NGG is four of them, and the CRISPR literature prefers some over others.
• Show flagged lists the guides with a ! too. They are hidden until it is pressed, so the list opens on the clean ones and the title's 139 of 345 says how many were set aside.

Spacer length is not a filter: it is fixed by the nuclease, so every guide in a list has the same one.

What it cannot tell you

Off-targets are counted in the open documents only. PlasmidPop has no genome and makes no network request to search one, so the counts say where else a guide would cut _this plasmid_ — which is what matters when the guide has to leave the backbone alone — and nothing whatever about specificity in a cell. Check a guide against the host genome with a genome-wide tool (CRISPOR, CHOPCHOP, Benchling) before ordering it. The panel says this under the options until Got it is pressed; from then on the dotted Off-targets up to label keeps it as a tooltip.

There is no on-target efficiency score. The published scores are trained models, and PlasmidPop ships a number only when it can be checked against the implementation it came from; until then the flags below are what it will stand behind.

Nucleases

Nuclease picks what to look for:

Custom PAM… takes a PAM written in IUPAC codes — N any base, R A or G, V not T, and so on — and a spacer length of 15–30. A custom PAM is taken to sit 3′ of the spacer and to cut bluntly three bases in from it, which is what the engineered Cas9 variants do (NG for Cas9-NG, NRN for SpRY, NNNRRT for SaCas9-KKH). A nuclease with another geometry needs a preset.

An ambiguous base in the document never makes a guide: a PAM has to be certain and a spacer has to be plain ACGT, so an N in a sequencing gap cannot be mistaken for a target. The same N does still count _against_ a guide as a possible off-target, so an ambiguity never hides one.

Off-targets

Off-targets up to _n_ mismatches sets how different another site may be and still be counted, from 0 to 4. The counts in a row read exact, then one mismatch, then two, and so on: 1 · 0 · 2 · 0 is one exact second site and two sites differing in two bases. Only places with a real PAM are counted — a near-match with no PAM is not a target.

Count off-targets in the other open documents too adds every other open nucleotide document to the search, which is how to check that a guide for an insert does not also cut the vector it is going into. The selected guide's Other sites list says which document each hit is in, and Show jumps to the ones in this one.

Sort by orders the list along the molecule, worst-first by off-targets — fewest exact hits, then fewest near ones — or by GC content, highest first, or by specificity, highest first (SpCas9 only; guides without a score go last).

Specificity score

For SpCas9 (NGG, 20 nt) a guide's details show a Specificity from 0 to 100: the MIT score of Hsu et al. (2013). Each off-target site is weighted by where its mismatches fall (mismatches near the PAM hurt most, and a few spread-out ones count less than a cluster), and 100 means no other site was found. Two exact copies of a guide score 50.

Read it as an upper bound. It sees only the sites within Off-targets up to _n_ mismatches and only the open documents, so a guide that scores 100 here is not known to be specific genome-wide. It is not an efficiency score, which PlasmidPop does not give. Other nucleases show none, because the weights were measured for SpCas9 only.

Flags

A guide is flagged for the things that are known to go wrong, none of which are efficiency predictions:

• Binds somewhere else exactly — the spacer occurs at another PAM site in what was searched.
• TTTT ends a U6 transcript — four T's in a row terminate Pol III, so a guide expressed from a U6 promoter is cut short.
• Low or high GC — outside 40–80%.
• A run of one base — five or more of the same base in a row.

Narrowing the scan

Only cuts in the selection keeps the guides whose cut falls inside the selected range — select a feature in the feature list, or a stretch in either view, to get the guides that cut there. The cut, not the protospacer, is what has to be in the range, so a guide reading into the region from outside it still counts if it cuts inside. For AsCas12a, whose two strands are cut five bases apart, it is the break on the PAM strand that counts.

The region is taken when you tick the box, and the panel shows it under the box. Clicking a guide afterwards selects that guide without narrowing the list to it. To narrow to a different stretch, select it and press Use the selection now, or untick and tick the box again.

Base editing

With SpCas9 (NGG), Base editor marks where a base editor can act. Choose CBE (C to T, protospacer positions 4–8), ABE7.10 (A to G, 4–7) or ABE8e (A to G, 4–8). Positions count from the spacer's 5′ end, so the PAM is 21–23. Each row then shows how many bases of the right kind sit in the window, Only with a C to edit (or A) hides the guides with none, and a guide's details list the bases (such as C4, C8), where they are, and the spacer with all of them converted. More than one base listed means bystander edits: the window does not say how likely each is. On a reverse-strand guide the base in the sequence is the complement, so a C to T edit is a G to A there. The windows are measured for a 20 nt SpCas9 spacer, so the choice is not offered for other nucleases.

Prime editing

The Prime editing group designs pegRNAs for the guides found above. Give the Edit at position (1-based), how many bases it Replaces and what it is replaced With (nothing deletes; zero bases replaced inserts), or select the bases and press Use the selection. PBS sets the primer binding site length, 8–17 (13 by default).

For each guide whose nick lies up to 30 bases before the edit on the guide's strand, nearest first, the list gives the primer binding site, the reverse-transcriptase template (the edit plus ten bases of homology) and the 3′ extension, which is the template followed by the primer binding site. Copy spacer and extension puts the two to order on the clipboard; add them to your scaffold. It also says whether the edit changes the PAM, which stops the edited site being nicked again, and flags a template that starts with C, a primer binding site outside 40–60% GC, and TTTT. An edit before the nick, or too far after it, has no pegRNA from that guide. PlasmidPop does not rank pegRNAs by predicted efficiency, and does not design the second nick of PE3.

Paired nickases

For SpCas9 and the other nucleases that cut both strands at one point, the Paired nickases group lists pairs of guides on opposite strands whose nicks, one on each strand, are close enough to make a staggered double-strand break. Each pair shows its two guides and the overhang: the offset between the nicks, and whether it is a 5′ or 3′ overhang. Nickase picks which strand each guide's nick falls on: D10A nicks the strand the guide pairs with, H840A the PAM strand. With D10A, guides whose PAMs face outward (PAM-out) leave 5′ overhangs, the arrangement reported to work best, and guides facing inward leave 3′ overhangs. Nicks at most … apart sets the largest offset listed (100 by default). Show selects the span of both guides. Pairs are ranked by the distance between the nicks only; no pair is scored, and nothing is checked for off-target nicking beyond what each guide's own row shows. Pairing works through the origin of a circular sequence.

Ordering and annotating a guide

The selected guide's box, under its row, gives its protospacer range, spacer, PAM, where both strands are cut, GC and flags, and the buttons below.

Add as feature annotates the protospacer on the document as a misc_feature on its strand, with the PAM in a note, and takes you to the feature list to name it. It is an ordinary edit, so undo takes it back.

Add N listed as features, above the list, annotates every guide the filters leave in the list in one go, so a whole set is saved with the file, exported to GenBank and drawn on the maps like any feature. Narrow the list first (a selection, the filters, hiding flagged guides) to annotate only the ones you want. One undo takes them all back.

Oligos to order writes the two oligos to anneal and clone, 5′→3′. pX330 / lentiCRISPRv2 adds the CACC and AAAC overhangs those vectors' BbsI and BsmBI sites take, and a G in front of a spacer that has not got one, since the U6 promoter starts with G. These are Cas9 sgRNA vectors, so the choice is offered for SpCas9, SaCas9 and custom PAMs but not for AsCas12a, whose crRNA vectors take other overhangs. No overhangs gives the bare spacer and its reverse complement, for any nuclease. Copy oligos puts both on the clipboard; Save to My primers puts them in My primers, where they can be exported with the rest of an order.