CALITAS

This repository is home to CALITAS, a CRISPR-Cas-aware ALigner for In silico off-TArget Search. CALITAS implements a customized gapped alignment of guide sequences to genomes and other reference sequences, returning consistent and non-redundant alignments.

Overview

CALITAS is a suite of bioinformatic tools for enumerating candidate off-target sites for CRISPR guide sequences and standardizing their alignment. It features tools for searching an entire genome for candidate off-target sites, as well as for aligning guides to specific sequences or locations in a genome.

Key features of CALITAS include:

1. Detection of all candidate off-target sites up to the requested number of mismatches and gaps
2. Elimination of redundant alignments resulting in a single best or canonical alignment per locus
3. Searches with multiple PAM sequences and/or PAM-less searching
4. Integration of known variants in VCF format into genome-wide off-target searches
5. Customizable scoring system for weighting mismatches vs. gaps and differences in the protospacer vs. PAM

CALITAS is not intended to predict active off-target sites, but rather to enumerate candidate off-target sites for further investigation.

Getting CALITAS

Releases

Binary releases of CALITAS are available from the releases page on GitHub. The downloadable JAR files contain CALITAS and all dependencies, and need only a Java RunTime version 8 or higher installed. Once downloaded CALITAS can be run as follows to produce usage information:

java -Xmx8g -jar calitas.jar

Bioconda

Relases of CALITAS are also available via Bioconda. To use these releases you will first need to install the conda package manager - Miniconda is recommended if you do not already use conda. CALITAS can then be installed with:

conda install -c bioconda calitas

The conda release comes with a small helper script and can be run simply as calitas.

Building from Source

Both release and development versions of the code can be built from source. Builds are performed with sbt. Once sbt is installed run:

sbt clean assembly

This will build a JAR equivalent to a release JAR at calitas/target/scala-2.12/calitas.jar

Publication

A CALITAS manuscript is currently in review. This section will be updated with a reference to the publication when it is available.

Usage

CALITAS has four available sub-commands:

• AlignToReference performs glocal alignment of query sequence to a window on the reference
• PairwiseAlignSequences performs pairwise alignment of sequences
• PrepareVcf prepares a VCF for optimal use by SearchReference
• SearchReference searches a FASTA file for alignments of a guide+PAM(s)

It should be noted that commands which use a genome or reference FASTA file require that the FASTA file have both an index and a sequence dictionary. These can be generated using samtools as follows:

samtools faidx ref.fa
samtools dict -a <assembly-name> -s <species> -o ref.dict ref.fa

The following is an example of invoking SearchReference to find candidate off target sequences in the HG38 genome for a single guide and PAM (note that location and sequence of the PAM is indicated by providing guide sequence in upper case and PAM sequence in lower case):

calitas SearchReference \
  -i CTTGCCCCACAGGGCAGTAAnrg \
  -I myguide \
  -r hg38.fa \
  -o myguide.hits.txt \
  --max-guide-diffs 5 \
  --max-pam-mismatches 1 \
  --max-gaps-between-guide-and-pam 3

The last three parameters are optional and replicate the defaults. Additional options are available; detailed usage including all available parameters can be obtained by running calitas SearchReference.

The following is an example of running AlignToReference to produce standarized alignments at locations where guide(s) are known to align. The invocation will produce the single best alignment per query sequence and target location:

calitas AlignToReference -i input.txt -r hg38.fa -o output.txt --window-size 60

With the following being an example of the tab-delimited input file for AlignToReference:

id    query    chrom    position
1    CTTGCCCCACAGGGCAGTAAnrg    chr1    13358
2    CTTGCCCCACAGGGCAGTAAnrg    chr1    510578
3    CTTGCCCCACAGGGCAGTAAnrg    chr1    844033

The output of both SearchReference and AlignToReference is a tab-delimited text file with one row per candidate off-target site including the following columns:

column name	description
guide_id	Name/ID of guide.
unpadded_guide_sequence	The sequence of the guide used, unpadded.
genome_build	The assembly name of the searched genome (e.g. HG38).
chromosome	Chromosome for target sequence alignment (eg: chr3).
coordinate_start	Start of the unpadded target sequence in the genome, 0-based open ended, excluding PAM.
coordinate_end	End of the unpadded target sequence in the genome, 0-based open ended, excluding PAM.
strand	Either + or -. The reported strand is the strand of the genome which matches the guide sequence. E.g. if strand is reported as + this means the guide resembles the sequence on the top strand of the genome, and will bind to the bottom strand of the genome.
unpadded_target_sequence	The unpadded target sequence (as DNA) as found in the genome, without gaps/bulges, excluding PAM. Reported sequence matches the reported strand (i.e. - strand hits will report the reverse complement of the genomic sequence).
ten_bases_5_prime	The 10 bases from the reference genome immediately 5’ of the off-target location (coordinate_start/coordinate_end), respecting strand.
ten_bases_3_prime	The 10 bases from the reference genome immediately 3’ of the off-target location (coordinate_start/coordinate_end), respecting strand.
pam_used	PAM used in the alignment (eg: nrg).
variant_id	When searching using a VCF, a semi-colon separated list of variant IDs (e.g. rs1234;rs2345) that have non-reference alleles present in the off-target alignment. May be empty when no variants are present.
variant_description	When searching using a VCF, a semi-colon separated list of variant descriptions in the format id:pos:ref>alt:af where pos is the position within the alignment and af is the allele frequency of the alternate allele.
variant_vcf	When searching using a VCF, a string composed of filname of the VCF followed by a colon (:) and then the MD5 of the VCF.
allele_frequency	When searching using a VCF, the minimum allele frequency of any variant included in the target alignment.
score	Alignment score (including PAM).
guide_mm	Mismatches in the guide region (excluding PAM).
guide_gaps	Gaps in the guide region (excluding PAM).
guide_mm_plus_gaps	Total gaps and mismatches in the guide region (excluding PAM).
pam_mm	Mismatches in the PAM region.
total_mm_plus_gaps	Total count of mismatches and gaps across the both guide and PAM regions.
padded_guide	Guide + PAM sequence with padding for mismatches and bulges/gaps.
padded_alignment	Visual representation of the guide-target alignment: `
padded_target	Target sequence, including PAM, with padding for mismatches and bulges/gaps.
padded_extra_8_bases_5_prime	an additional 8 bases on the 5’ side of the padded_target.
padded_extra_8_bases_3_prime	an additional 8 bases on the 3’ side of the padded_target.
cigar	Cigar representation of guide sequence alignment.
unpadded_guide_sequence_length	Length of guide sequence not including PAM, nor gaps/bulges.
unpadded_target_sequence_length	Length of target sequence not including PAM, nor gaps/bulges.
aligner	Aligner name, e.g. CALITAS:SearchReference.
aligner_version	Version number of the CALITAS software used to produce the alignments.
aligner_search_pam	Comma-separated list of pams used during the search.
aligner_other_parameters	A semicolon-separated list of parameters provided when running CALITAS.
time_stamp	A date and time stamp for when the alignment run was started (UTC) in this format: Wed Jan 6 16:58:29 UTC 2021.

Note on coordinates produced by CALITAS

All genome coordinates in CALITAS output files are 0-based open ended, the same systemm used in BED files. I.e. the first base of a chromosome or sequence is represented by 0, and when describing a region on chromosome we specify the first base included in the interval as the start and the base after the interval as the end. E.g. the first 10bp on a chromosome would be represented as 0-10.