Software requirements

Installation from Bioconductor

crisprViz can be installed from Bioconductor using the following commands in a fresh R session:

install.packages("BiocManager")
BiocManager::install("crisprViz")

Visualizing the best gRNAs for a given gene

Suppose we want to design the four best gRNAs using the SpCas9 CRISPR nuclease to knockout the human KRAS gene. To have the greatest impact on gene function we want to prioritize gRNAs that have greater isoform coverage, and target closer to the 5’ end of the CDS.

Let’s load a precomputed GuideSet object containing all possible gRNAs targeting the the CDS of KRAS, and a GRangesList object describing the gene model for KRAS.

data("krasGuideSet", package="crisprViz")
data("krasGeneModel", package="crisprViz")
length(krasGuideSet) # number of candidate gRNAs

## [1] 52

For how to design such gRNAs, see the crisprDesign package. Before we plot all of our candidate gRNAs, let’s first generate a simple plot with a few gRNAs to familiarize ourselves with some plot components and options.

plotGuideSet(krasGuideSet[1:4],
             geneModel=krasGeneModel,
             targetGene="KRAS")

There are a few things to note here.

• The ideogram track and genome axis track are at the top of our plot and give us coordinate information.
• Our targetGene KRAS is plotted next, using coordinates from the provided gene model krasGeneModel, followed by our spacer subset. The name of each track is given on the left.
• The strand information for each track is included in the label: <- for reverse strand and -> for forward strand.
• While we can identify which exon each spacer targets (which may be sufficient), the plot window is too large to provide further information.
• The plot only shows the 3’ end of KRAS, rather than the entire gene.

This last point is important: the default plot window is set by the spacers’ ranges in the input GuideSet object. We can manually adjust this window by using the from, to, extend.left, and extend.right arguments. Here is the same plot adjusted to show the whole KRAS gene, which also reveals an additional isoform that is not targeted by any spacer in this example subset.

from <- min(start(krasGeneModel$transcripts))
to <- max(end(krasGeneModel$transcripts))
plotGuideSet(krasGuideSet[1:4],
             geneModel=krasGeneModel,
             targetGene="KRAS",
             from=from,
             to=to,
             extend.left=1000,
             extend.right=1000)

As calculated above, there are a total of 52 candidate gRNAs targeting the CDS of KRAS. Including all of them could crowd the plot space, making it difficult to interpret. To alleviate this we can hide the gRNA labels by setting the showGuideLabels argument to FALSE.

plotGuideSet(krasGuideSet,
             geneModel=krasGeneModel,
             targetGene="KRAS",
             showGuideLabels=FALSE,
             from=from,
             to=to,
             extend.left=1000,
             extend.right=1000)

At the gene level, the plot window is too large to discern details for each spacer target. However, we can see five distinct clusters of spacer target locations that cover the CDS of KRAS. The spacers in the 5’-most cluster (on the reverse strand) target the only coding region of the gene that is expressed by all isoforms, making it an ideal target for our scenario.

We can see which gRNAs target this region by returning showGuideLabels to its default value of TRUE, and by adjusting the plot window to focus on our exon of interest.

# new window range around target exon
targetExon <- queryTxObject(krasGeneModel,
                            featureType="cds",
                            queryColumn="exon_id",
                            queryValue="ENSE00000936617")
targetExon <- unique(targetExon)
from <- start(targetExon)
to <- end(targetExon)
plotGuideSet(krasGuideSet,
             geneModel=krasGeneModel,
             targetGene="KRAS",
             from=from,
             to=to,
             extend.left=20,
             extend.right=20)

At this resolution we can get a much better idea of spacer location and orientation. In particular, the PAM sequence is visible as a narrow box on the 3’ side of our protospacer sequences. We can also distinctly see which spacer targets overlap each other–it may be best to avoid pairing such spacers in some applications lest they sterically interfere with each other.

If we have many gRNA targets in a smaller window and are not concerned with overlaps, we can configure the plot to only show the pam_site, rather than the entire protospacer and PAM sequence, by setting pamSiteOnly to TRUE.

plotGuideSet(krasGuideSet,
             geneModel=krasGeneModel,
             targetGene="KRAS",
             from=from,
             to=to,
             extend.left=20,
             extend.right=20,
             pamSiteOnly=TRUE)

Let’s filter our GuideSet by the spacer names in the plot then pass an on-target score column in our GuideSet to onTargetScores to color the spacers according to that score, with darker blue colors indicating higher scores. Note that for this plot we need not provide values for from and to, as the plot window adjusts to our filtered GuideSet.

selectedGuides <- c("spacer_80", "spacer_84", "spacer_88", "spacer_92",
                    "spacer_96", "spacer_100", "spacer_104", "spacer_108",
                    "spacer_112")
candidateGuides <- krasGuideSet[selectedGuides]
plotGuideSet(candidateGuides,
             geneModel=krasGeneModel,
             targetGene="KRAS",
             onTargetScore="score_deephf")

Plotting for precision targeting

For a given CRISPR application, the target region may consist of only several base pairs rather than an exon or entire gene CDS. In these instances it may be important to know exactly where the gRNAs target, and plots of gRNAs must be at a resolution capable of distinguishing individual bases. This is often the case for CRISPR base editor (CRISPRbe) applications, as the editing window for each gRNA is narrow and the results are specific to each target sequence.

In this example, we will zoom in on a few gRNAs targeting the 5’ end of the human GPR21 gene. We want our plot to include genomic sequence information so we will set the bsgenome argument to the same BSgenome object we used to create our GuideSet.

First, we load the precomputed GuideSet and gene model objects for GPR21,

data("gpr21GuideSet", package="crisprViz")
data("gpr21GeneModel", package="crisprViz")

and then plot the gRNAs.

plotGuideSet(gpr21GuideSet,
             geneModel=gpr21GeneModel,
             targetGene="GPR21",
             bsgenome=BSgenome.Hsapiens.UCSC.hg38,
             margin=0.3)

The genomic sequence is given at the bottom of the plot as color-coded boxes. The color scheme for describing the nucleotides is given in the biovizBase package. If the plot has sufficient space, it will display nucleotide symbols rather than boxes. We can accomplish this by plotting a narrower range or by increasing the width of our plot space (see “Setting plot size” section).

The plot above was generated with a plot space width of 6 inches; here’s the same plot after we increase the width to 10 inches:

# increase plot width from 6" to 10"
plotGuideSet(gpr21GuideSet,
             geneModel=gpr21GeneModel,
             targetGene="GPR21",
             bsgenome=BSgenome.Hsapiens.UCSC.hg38,
             margin=0.3)

CRISPRa and adding genomic annotations

In this scenario we want to increase expression of the human MMP7 gene via CRISPR activation (CRISPRa). We will use the SpCas9 CRISPR nuclease.

data("mmp7GuideSet", package="crisprViz")
data("mmp7GeneModel", package="crisprViz")

The GuideSet contains candidate gRNAs in the 2kb window immediately upstream of the TSS of MMP7. We will also use a GRanges object containing repeat elements in this region:

data("repeats", package="crisprViz")

Let’s begin by plotting our GuideSet, and adding a track of repeat elements using the annotations argument. Our guideSet also contains SNP annotation, which we would also prefer our gRNAs to not overlap. To include a SNP annotation track, we will set includeSNPTrack=TRUE (default).

from <- min(start(mmp7GuideSet))
to <- max(end(mmp7GuideSet))
plotGuideSet(mmp7GuideSet,
             geneModel=mmp7GeneModel,
             targetGene="MMP7",
             guideStacking="dense",
             annotations=list(Repeats=repeats),
             pamSiteOnly=TRUE,
             from=from,
             to=to,
             extend.left=600,
             extend.right=100,
             includeSNPTrack=TRUE)

Some of our candidate gRNAs target repeat elements and likely target a large number of loci in the genome, potentially causing unintended effects, or overlap with SNPs, which can reduce its activity. Let’s remove these gRNAs and regenerate the plot.

filteredGuideSet <- crisprDesign::removeRepeats(mmp7GuideSet,
                                                gr.repeats=repeats)
filteredGuideSet <- filteredGuideSet[!filteredGuideSet$hasSNP]
plotGuideSet(filteredGuideSet,
             geneModel=mmp7GeneModel,
             targetGene="MMP7",
             guideStacking="dense",
             annotations=list(Repeats=repeats),
             pamSiteOnly=TRUE,
             from=from,
             to=to,
             extend.left=600,
             extend.right=100,
             includeSNPTrack=TRUE)

Note how removing gRNAs that overlap SNPs from our GuideSet also removed the SNP track. To prevent plotting an empty track, plotGuideSet will only include a SNPs track if at least one gRNA includes SNP annotation (i.e. overlaps a SNP).

Conversely, there are specific genomic regions that would be beneficial to target, such as CAGE peaks and DNase I Hypersensitivity tracks. We show in the inst\scripts folder how to obtain such data from the Bioconductor package AnnotationHub, but for the sake of time, we have precomputed those objects and they can be loaded from the crisprViz package directly:

data("cage", package="crisprViz")
data("dnase", package="crisprViz")

We now plot gRNAs alongside with those two tracks:

plotGuideSet(filteredGuideSet,
             geneModel=mmp7GeneModel,
             targetGene="MMP7",
             guideStacking="dense",
             annotations=list(CAGE=cage, DNase=dnase),
             pamSiteOnly=TRUE,
             from=from,
             to=to,
             extend.left=600,
             extend.right=100)

Let’s filter our GuideSet for guides overlapping the plotted DNase site then regenerate the plot.

# filter GuideSet for gRNAs overlapping DNase track
overlaps <- findOverlaps(filteredGuideSet, dnase, ignore.strand=TRUE)
finalGuideSet <- filteredGuideSet[queryHits(overlaps)]
plotGuideSet(finalGuideSet,
             geneModel=mmp7GeneModel,
             targetGene="MMP7",
             guideStacking="dense",
             annotations=list(CAGE=cage, DNase=dnase),
             pamSiteOnly=TRUE,
             margin=0.4)

Comparing multiple GuideSets targeting the same region

The choice of the CRISPR nuclease can be influenced by the abundance of PAM sequences recognized by a given nuclease in the target region. For example, we would expect AT-rich regions to have fewer possible targets for the SpCas9 nuclease, whose PAM is NGG. In these regions, the CRISPR nuclease AsCas12a, whose PAM is TTTV, may prove more appropriate. Given multiple GuideSets targeting the same region, we can compare the gRNAs of each in the same plot using plotMultipleGuideSets.

Here, we pass our GuideSets targeting an exon in the human gene LTN1 in a named list. Note that there are no available options for displaying guide labels or guide stacking, and only the PAM sites are plotted. We will also add a track to monitor the percent GC content (using a window roughly the length of our protospacers). Not surprisingly, this AT-rich region has fewer targets for SpCas9 compared to AsCas12a. (Note: when plotting several GuideSets you may need to increase the height of the plot space in order for the track names to appear on the left side; see “Setting plot size” below.)

We first load the precomputed GuideSet objects:

data("cas9GuideSet", package="crisprViz")
data("cas12aGuideSet", package="crisprViz")
data("ltn1GeneModel", package="crisprViz")

plotMultipleGuideSets(list(SpCas9=cas9GuideSet, AsCas12a=cas12aGuideSet),
                      geneModel=ltn1GeneModel,
                      targetGene="LTN1",
                      bsgenome=BSgenome.Hsapiens.UCSC.hg38,
                      margin=0.2,
                      gcWindow=10)