CRISPRome: A Comprehensive CRISPR-Cas Resource Database

CRISPRome is a large publicly available CRISPR-Cas system database to date, systematically characterizing CRISPR-Cas elements across a diverse set of bacterial and archaeal genomes. This resource serves as a foundation for studying the global distribution and characteristics of CRISPR systems in microbes.

Data Collection

To construct CRISPRome, we compiled data from multiple sources:

RefGen Database:

1,414,360 bacterial genomes

13,258 archaeal genomes

Extracted from GenBank and RefSeq

Metagenomic Datasets:

26,565 genomes/metagenome-assembled genomes (MAGs)

Collected from three human body sites and three natural environments

CRISPR-Cas System Identification

We identified CRISPR-Cas systems using CRISPRCasFinder v2.0.355, extracting both spacers and CRISPR arrays:

Spacers:

21,282,263 total spacers

14,968,498 (70%) associated with Cas proteins

CRISPR Arrays:

3,937,928 total arrays

1,429,909 (36%) linked to Cas proteins

Spacer-Target Identification Process

Spacer Deduplication & Alignment

Spacers were deduplicated and aligned using Blastn-short to a database of 7,042,467 genomes from RefSeq, GenBank, ICEberg, IMG/VR, and PLSDB.

makeblastdb was used to convert genome sequences into searchable databases.

Blastn Alignment & Optimization

Blastn was run with optimized parameters (word_size 11, qcov_hsp_perc 95, perc_identity 95, max_hsps 3) to balance speed and accuracy.

High-performance computing clusters managed by Slurm were used for large-scale processing.

Filtering & Prioritization

Hits were sorted by bitscore, keeping the top 10% for accuracy.

Prioritization order: viruses > plasmids = ICE > other sources, reflecting CRISPR-Cas immunity targets.

Removing CRISPR Array Artifacts

Alignments overlapping known CRISPR array regions were removed to avoid false positives.

Additional CRISPR arrays were identified from NCBI genome annotations to refine filtering.

Detecting Potential CRISPR Arrays

Arrays were defined by detecting regularly interspaced protospacers.

A threshold of ≥4 protospacers was used to minimize false positives, leading to the exclusion of 16,777 targets.

LCA-Based One-to-One Spacer-Target Assignment

TaxonKit lca was used to determine the Lowest Common Ancestor (LCA) of multiple target matches.

This ensured each spacer was assigned to a single, biologically relevant target.