This application takes a partial prokaryotic virus sequence and predicts the host. It utilizes the fact that foreign DNA molecules such as phage genomes, plasmid and mobile genetic elements, can be included as spacers in CRISPR arrays in bacteria and archaea. Therefore, spacers in CRISPR-arrays provide an indication of what hosts have defended against the virus, or a similar virus. FASTA-formatted BLAST databases of CRISPR spacers from bacteria and archaea (predicted by CRISPRDetect 2.4) were built. In the database, each spacer sequence was linked to the genomic and taxonomic information of source organisms in the headers. When DNA sequences are provided as queries by the user, BLASTN will be used to find any spacer targets in each sequence, and the genomic and taxonomical information of their owners can be obtained.
Validation data, and R scripts used for PPV calculations and testing statistics are in this GitHub repo: http://github.com/davidchyou/CRISPRHost_data_analysis. By considering spacer matches with an e-value no higher than 1e-8 and at least 95% of all nucleotides in spacer-target hybrids to be matched, the PPV calculations (by exact name-matching) are the following.
| PPV-Order | PPV-Family | PPV-Genus | |
|---|---|---|---|
| Bacteria | 0.97 | 0.96 | 0.92 |
| Archaea | 0.98 | 0.98 | 0.7 |
A web version is available through a Galaxy interface here: http://crispr.otago.ac.nz:8080 (under "Virus: Classification").
The source directory was first uploaded to GitHub in 2018, and was presented by Chris Brown at CRISPR2019, Molecular Biology of Archaea 2019, and Asian Transcription Conference 2019.
A manuscript is in preparation (Chyou, Biswas, Fineran, Brown).
As a minimum, nucleic-acid sequences need to be provided as a FASTA or a multi-FASTA formatted file, along with the path to the output directory. Additional options (next section) can also be specified. CRISPRHost can be called outside the source directory.
perl CRISPRHost.pl -in <fasta_path> -out <dir_path> <options>
perl /path/CRISPRHost.pl -in <fasta_path> -out <dir_path> <options>
The full list of command-line options is displayed below.
-in <path> [Required] A FASTA file of viral genome, mobile genetic elements, or sequences.
-out <path> [Required] Path to the output directory.
-kingdom <A|B|AB> Select a kingdom (A=archaea, B=bacteria, AB=archaea+bacteria).
The corresponding BLASTDB of CRISPR spacers (non-redundant) will then be used:
1. A: RefSeq95 archaeal, DB/CRISPRBankSpacers_4_95_2555_100_archaea_refseq_nr.fa
2. B: RefSeq95 bacterial, DB/CRISPRBankSpacers_4_95_2555_100_bacteria_refseq_nr.fa
3. AB: RedSeq95 archaeal+bacterial, DB/CRISPRBankSpacers_4_95_2555_100_all_refseq_nr.fa
During the first run, the app will unzip "DB.zip" for the DB directory.
Default: AB.
-mask_arrays Optionally run MINCED to predict predict arrays and mask them using BEDTOOLS.
-dbsize <integer> Number of nucleotides in BLASTDB, optional parameter for BLASTN. Default: 10000.
-e_cutoff <numerical> E-value cutoff for BLASTN spacer hits. Default: 0.001.
-b_cutoff <numerical> Bit-score cutoff for BLASTN spacer hits. Default: 50.
-pr_cutoff <numerical 0-1> Minimum proportion of base-paired nucleotides in spacer-target hybrids. Default: 0.7.
-min_nocsh <integer> Minimum number of CRISPR spacer hits per sequence. Default: 1.
-max_target_seqs <integer> Maximum number of hits to be returned by BLASTN for post-processing. Default: 500.
-h Show this help.
-help Show this help (same as -h).
The following output files are generated.
summary.csv Files showing the key results. The columns are:
1. Spacer hit statistics
VIRAL_GENOME Viral genome identifier.
N_HOST_SPACER Number of hits from spacers of the same sequence.
E_VALUE BLASTN e-value.
BITSCORE BLASTN bitscore.
PROP_NT_MATCHED_IN_HYBRID Proportion of matched nucleotide in a spacer-target hybrid.
2. Attributes of matched spacers
SPACER_ID Spacer identifier.
KNOWN_SPECIES A list of species where the same spacer can be found.
These species are the predicted hosts.
TARGET_ALIGN_START Start-coordinate of the target alignment.
SEQ_TARGET_ALIGNED Sequence of the spacer target.
SPACER_ORI_ALIGN Whether the match is corresponding to the reverse-complement
of the spacer (minus) or not (plus).
full_results.csv Files showing the key results in summary.csv, with additional information and statistics
included.
spacer_matched.fna A list of spacers matched in multi-FASTA format.
c_arrays.gff Any CRISPR arrays found by MINCED and masked by BEDTOOLS, if -mask_arrays is used
CRISPRHost requires the following dependencies. The executables of applications are provided, but BioPerl, R and Rscript will need to be installed separately.
- NCBI BLAST suite: blastn, makeblastdb and blastdbcmd (provided)
- Bedtools (provided)
- minced (provided)
- Perl library BioPerl
- R 3.6+ (no add-on libraries needed) and Rscript
Example commands:
perl CRISPRHost.pl -in NC_034623.fna -out test_NC_034623
perl CRISPRHost.pl -in NC_034623.fna -out test_NC_034623 -kingdom AB
perl CRISPRHost.pl -in NC_034623.fna -out test_NC_034623 -kingdom AB -mask_arrays