Using DeepSpCas9variants

Using DeepSpCas9variants Model_Overview

DeepSpCas9variants designs guide RNAs and predicts the efficiency of CRISPR–Cas9 (SpCas9) and various SpCas9 variants with optimized scaffold.It supports:

SpCas9

Sniper-Cas9

xCas9

eSpCas9 (1.1)

SpCas9-HF1

HypaCas9

evoCas9

VRQR variant

SpCas9-NG

Note: DeepCas9variants (select DeepCas9variants in the Comparison Mode within DeepBE) provides guide RNA designs and efficiency predictions for six additional Cas9 variants. DeepSpCas9, DeepSpCas9-NG, and DeepxCas9 provide guide RNA designs and efficiency predictions for SpCas9, SpCas9-NG, and xCas9 with conventional scaffold.Load Example

The Load Example button is able to replicate the example inputs shown in this help page for quick testing of the DeepSpCas9variants web server. Simply click the Load Example button then press Submit Job to obtain an example results page.Input Input_Text

Input Format: Nine formats reach the same designer, and each one resolves to the target sequences this model scores: FASTA, Gene Symbol, NMID, Ensembl ID, HGNC ID, Genomic Position, Edit Sequence (an unedited/edited pair), ClinVar ID and COSMIC ID. Picking one shows the fields it needs and hides the rest. Edit Sequence, ClinVar ID and COSMIC ID are the three formats that specify an edit, so those three also produce the homology-directed repair (HDR) donor columns described below.Input Method (FASTA): You can either enter the FASTA text or upload a FASTA file.

Enter FASTA Text: Enter target sequences in FASTA format. The sequence length must be >= 30 bp.

Upload FASTA File: Upload a .txt, .fa, or .fasta file containing the target sequences.

When uploading files, press the [+] button to add files. Uploaded files can be cleared by pressing either of the checkmark buttons.Example Input:>MIR99AHG/LINC00478 (chr21:16,587,284-16,587,784) CAGACAGCATTAGAGTTGGCTTGAGAATTGCCGTACTTTGCTTCCCTTTGTATGTATTTC TTGTATGCTGCCGAGTCACTGATGGCTAGCTCTGTCTGGCAAGTAATTCAAAAATGCTGT TTATGTAGAAAGGAAAGGTAGGGACTTTACCACACTCTGTCATTAAAGGGAGCAATTGAA GAACAAAGGAACTGAGTAAATACCTATATATTGCCTTTTGTGTTGCGAAACACTGTAGCA CAAACACATTTGTGTTCAGCCAAATGTTTTACTTCCTTTTGTAATAACGCATATAGTAGG TTGTCTCCACATATGTACAAGAATCCATATTTTATTTAAACGTATATAGTCAATTGTTCA TATTTATAGGCTGCAAACATTTCTCAATCTCAAAGACTTTTACATATCCACTCCCACACA GCTATTTGTTATTATTTTAAAAGTTCTTAAATTAAAAAAAAAAATAAAATATACTAATAT CTCTGTTGGTTGATTTTATTASpeciesA gene identifier is looked up in Human (GENCODE v50 on GRCh38) or Mouse (GENCODE M39 on GRCm39). The selector appears with the four gene formats, and it is offered wherever the genome-wide off-target search offers Mouse, so a run that screens against the mouse genome can design in mouse as well.A mouse symbol from an older nomenclature resolves to the current one and the job log states the substitution — Olfr1 designs against Or1e16. Whichever of the four gene formats is chosen, Mouse replaces the identifier box with Enter Mouse Gene Name or ID, which takes a mouse symbol or a mouse Ensembl gene ID. ClinVar ID and COSMIC ID carry no species selector: both are human variant databases, and both resolve against the human reference.Homology-Directed Repair (HDR) Donor TemplatesThere is no HDR mode switch: the input format decides. When the input format specifies an edit — an unedited/edited sequence pair, a ClinVar ID or a COSMIC ID — every candidate guide is also reported with what it needs for HDR. The other six formats identify the target without specifying an edit, so no donor is designed from them and the donor columns are absent rather than blank.Cut site (1-based in the target sequence) is the blunt SpCas9-family cut, between protospacer positions 17 and 18 — three bases from the PAM.Each candidate carries these columns, in the results table and in the downloadable files alike:

Cut-to-edit distance (nt) — how far the break sits from the base being changed. Candidates are ranked by it (HDR donor rank in the target), because repair efficiency falls away with distance; candidates at the same distance keep the order the model listed them in.

Donor (sense) / Donor (antisense) — the single-stranded donor carrying the edit, in the orientation(s) you asked for, ready to order.

Donor length (nt), Left arm (nt), Right arm (nt), Requested arms (nt) — arms are measured outward from the union of the cut site and the edit, not from either alone, so that a guide cutting some distance away still spans both. The realised arms therefore differ from the requested pair whenever the cut sits outside the edit, and both are reported.

Indel length not a multiple of 3 — a length test on the edit itself. It is deliberately not called frame-disrupting: the donor is built from a sequence window, so whether the position is coding at all is not knowable here.

Blocking mutation — off by default, and - on every row when it is on. A substitution counts as blocking only once the changed PAM matches none of the PAMs the editor accepts. A row here may belong to any of the nine variants, so this model's PAM slot is searched unconstrained, as NNNN — which every four-base sequence satisfies — and no substitution can stop it matching. Donor note carries that reason. The control is offered for uniformity with the CRISPR–Cas web servers whose defined PAM leaves a base to change.

Donor note — what shaped this donor: why a blocking mutation was or was not made, and by how much the arms were extended outward past the requested lengths so that the donor spans the cut as well as the edit. Where a row carries no donor, the note gives the reason in its place — arms that do not fit inside the supplied window, naming the extra sequence or the shorter arm that would make them fit, or a window that could not be tied to exactly one input record.

What you can set: homology arm lengths (100/100 nt by default, 50/50, or a custom pair between 25 and 1000 nt), the donor strand, whether to add a blocking mutation, and the donor template type — a chemically synthesized single-stranded oligodeoxynucleotide (ssODN) by default, with the donor note saying when a template outgrows standard oligo synthesis and should be ordered as a long single-stranded DNA (ssDNA) fragment instead; long ssDNA for those longer templates; or a double-stranded DNA (dsDNA) duplex, which arrives as one Duplex donor (dsDNA) column instead of the strand pair. The guide sequence is deliberately not yours to specify — the guides are this model's own ranked candidates, and a donor is built for each of them.Guide Cloning OligosCloning Vector decides whether the results carry an ordering-ready oligo pair, and it is set to SpCas9 / lentiGuide-Puro (BbsI or BsmBI) by default. Every candidate then gains Oligo (forward) and Oligo (reverse), in the results table and in the downloadable files alike: the two strands that anneal into the digested vector, each presenting its own 4 nt 5' overhang — CACC on the forward strand, AAAC on the reverse. The guide between them is written (G/g)N19, position 1 replaced by the G the U6 promoter transcribes from.Setting the vector to None (No Oligo Output) leaves both columns out rather than blank. The four overhang fields — Forward 5′, Reverse 5′, Forward 3′ and Reverse 3′ — override what the vector supplies; left blank, each takes the vector's own. A guide carrying anything outside A, C, G and T gives an empty pair, because there is nothing orderable to write.Off-Target ScoreThe Off-target score (higher = less specific) column is a mismatch-tolerance score, not a count of sites in a genome. It asks a narrower question than its name suggests: for this guide, how much activity would survive a single mismatch, averaged over every single-mismatch variant of the protospacer. A guide that tolerates mismatches well is a guide that is more likely to cut somewhere it was not aimed at, which is why a higher number is the worse one.It comes from the SpCas9 cutting frequency determination (CFD) table of Doench, Fusi et al. (2016), so it needs no genome and costs nothing beyond the on-target prediction. The table is indexed over the 20 nt protospacer and takes no PAM, so it reads the spacer and nothing else: one number for the whole row, beside the sixteen activity scores that are per variant, and a wild-type SpCas9 measurement.Against the nine variants that number means three different things. For xCas9, the VRQR variant and SpCas9-NG it is the RNA-DNA duplex readout they share with wild type; what distinguishes those three sits in the PAM-interacting domain, which the table does not read. For eSpCas9 (1.1), SpCas9-HF1, HypaCas9 and evoCas9 it is an upper bound: those four are engineered to tolerate less mismatch than wild type, so the wild-type number overstates their tolerance. For Sniper-Cas9, DeepSniper carries an off-target model trained on that nuclease, which is the better readout wherever you can use it.It cannot tell you where an off-target is, and it does not know whether any of those variants occurs in your genome at all.For that, set Genome-wide off-target search to Human or Mouse. That search adds one Genome-wide specificity (higher = more specific) column, which runs the other way round: there, higher is better. Every site it finds is condensed into that single number; the individual loci are not listed, so no chromosome, coordinate or strand for them reaches the table. The two are different measurements and both may appear. The sites that search finds are scored on the same table, so the two columns differ only in the question they answer -- one guide against every single-mismatch variant of itself, one guide against the sites a genome really holds -- and never in how a mismatch is weighed.Genome-search scoring is where that last clause stops holding. The CFD table knows which base was swapped for which, and was fitted on SpCas9 with an NGG PAM; the MIT position weights (Hsu et al. 2013) know only how tolerant each position is, and were fitted across nucleases rather than on one. Neither is a superset of the other. Leaving the setting on CFD only is the behaviour described above; choosing CFD and MIT adds a second Genome-wide specificity, MIT (higher = more specific) column, aggregated from the same search over the same sites, so the two readings can be compared directly on the guides you are choosing between. The choice applies to the genome search alone -- averaged over a complete single-mismatch panel the MIT weights return one number for every guide of a given length, which is a constant rather than a measurement, so the genome-free column is always the table's.ResultsThe tool provides prediction scores for multiple SpCas9 variants. Results_1

The results page produces the guide RNA sequence, and the DeepSpCas9variants Score (%) for each designed guide RNA sequence.Prediction scores are divided into (G/g)N19 (using the U6 promoter, which requires a guanosine (G) as the first nucleotide to initiate transcription efficiently) and tRNA-N20 (using the endogenous tRNA-processing system) scores.Seven of the nine variants carry both scores: SpCas9, Sniper-Cas9, xCas9, eSpCas9 (1.1), SpCas9-HF1, HypaCas9 and evoCas9. The VRQR variant and SpCas9-NG carry one score each, so the results hold sixteen activity columns rather than eighteen, and the Selected Target Details panel groups those last two together.You may click on each row to update the Selected Target Details visualization. The red highlight is the PAM sequence.The general specificities (fidelity) of SpCas9 variants: evoCas9 (the highest specificity) ≫ HypaCas9 ≥ SpCas9-HF1 ≈ eSpCas9(1.1) > xCas9 > Sniper-Cas9 > SpCas9. Download_Results

You can download the full results as a ZIP file using the Download Results button.Details for the general specificity of SpCas9 variants GeneralSpecificity

Figure. Specificity of SpCas9 variants against the mismatches between the guide sequence of the single guide RNA (sgRNA) and target sequence. (a) A heat map showing specificity of SpCas9 variants against single-base mismatches between the guide RNA and target sequence. The specificity was calculated as 1 - (indel frequencies at mismatched target sequences divided by those at perfectly matched targets). The general specificity represents the average of specificity at all 20 positions. n = 21 - 24 target sequences for each position. (b) Effects of mismatch types on activities of SpCas9 variants. Relative indel frequencies at targets with different types of one-base mismatches are shown. The boxes represent the 25th, 50th, and 75th percentiles; whiskers show 10th and 90th percentiles. n = 76 - 79, 76 - 79, and 304 - 306 target sequences for wobble transition, non-wobble transition, and transversion mismatches, respectively. (c, d) Effect of the number of mismatched bases on the activities of SpCas9 variants. Relative indel frequencies at two- (c) and three- (d) base mismatched target sequences are shown. n = 138 - 139 (c) and 133 – 135 (d) target sequences. The boxes represent the 25th, 50th, and 75th percentiles; whiskers show 10th and 90th percentiles.Sample OutputsThe following example interactive outputs are available for reference:FASTA sequenceGene symbolClinVar variantSequence pairGenomic coordinatesData and LicensingCOSMIC. COSMIC is a product of Genome Research Limited (Wellcome Sanger Institute). It is free for academic and non-commercial use; commercial use requires a COSMIC licence. Lookups here resolve a single user-typed identifier — no COSMIC data is listed, browsable or served in bulk.ContactIf you have any questions, bug reports, or suggestions, please do not hesitate to contact us through the Support page.

DeepSpCas9variants web server

All web servers

Your browser does not support ES modules.
Please use a modern browser.