RNA-guided nucleases from CRISPR-Cas systems expand opportunities for precise, targeted genome modification. Endogenous CRISPR-Cas systems in many prokaryotes are attractive to circumvent expression, functionality, and unintended activity hurdles posed by heterologous CRISPR-Cas effectors. However, each CRISPR-Cas system recognizes a unique set of protospacer adjacent motifs (PAMs), which requires identification by extensive screening of randomized DNA libraries. This challenge hinders development of endogenous CRISPR-Cas systems, especially those based on multi-protein effectors and in organisms that are slow-growing or have transformation idiosyncrasies. To address this challenge, we present Spacer2PAM, an easy-to-use, easy-to-interpret R package built to predict and guide experimental determination of functional PAM sequences for any CRISPR-Cas system given its corresponding CRISPR array as input. Spacer2PAM can be used in a ‘Quick’ method to generate a single PAM prediction or in a ‘Comprehensive’ method to inform targeted PAM libraries small enough to screen in difficult to transform organisms. We demonstrate Spacer2PAM by predicting PAM sequences for industrially relevant organisms and experimentally identifying seven PAM sequences that mediate interference from the Spacer2PAM-informed PAM library for the type I-B CRISPR-Cas system from _Clostridium autoethanogenum_. We anticipate that Spacer2PAM will facilitate the use of endogenous CRISPR-Cas systems for industrial biotechnology and synthetic biology.

Spacer2PAM: A computational framework to guide experimental determination of functional CRISPR-Cas system PAM sequences

### Prediction of PAM sequences

All CRISPR arrays were retrieved from CRISPRCasdb, part of CRISPR-Cas++, which can be found at https://crisprcas.i2bc.paris-saclay.fr/ (28). Alignment of CRISPR array spacers to genomes was done via BLAST (29) either programmatically using Spacer2PAM or manually through the web interface. The BLASTn algorithm was used and Eukaryotes (taxid:2759) were excluded from the search database. All other manipulations of sequence information and prediction of PAM sequences were completed using Spacer2PAM which is available at https://github.com/grybnicky/Spacer2PAM. Spacer2PAM requires the following dependencies: dplyr, ggplot2, ggseqlogo (30), taxonomizr, HelpersMG, httr, jsonlite, spatstat.utils, readr and seqinr. Prophage prediction uses the Phaster API (31).

Briefly, Spacer2PAM can be used by passing the CRISPR-Cas system's host organism name and a user-defined identifier to setCRISPRInfo, which sets the name of the CRISPR-Cas system and defines file output names. The user then chooses one of two options to input the CRISPR array spacer sequence data. If starting with a FASTA file containing each spacer as an individual sequence, the user may call FASTA2DF to arrange the spacer sequences and other user input information about the CRISPR spacers (array number, length of each array, direction of each array, and the array consensus repeat sequence) into a dataframe which is suitable for downstream analysis with Spacer2PAM. We recommend that the user then call DF2FASTA to generate another FASTA file containing all the spacer sequences. Although the user already supplied a FASTA file with the same sequence information, doing so ensures that the title of each sequence is compatible with downstream Spacer2PAM functions. Alternatively, a user may start with a formatted dataframe containing the headers ‘Strain’, ‘Spacers’, ‘Array.Orientation’, ‘Repeat’, ‘Array’, and ‘Spacer’ and pass it to DF2FASTA to generate a FASTA file containing the spacer sequences with the appropriate labels.

Next, the user then submits the sequences from the FASTA file for alignment to BLAST. This step can either be done programmatically or manually. To send a query to the BLAST server and retrieve the result, call FASTA2Alignment and pass the file location of the properly formatted FASTA file generated from DF2FASTA. While we recommend this method, some CRISPR-Cas systems may contain too many spacers and exceed the query length limit for the NCBI BLAST API. In this instance, FASTA2Alignment will return an error message and encourage the user to visit the BLAST web interface. If using the web interface, select the BLASTn algorithm and to exclude Eukaryotes (taxid:2759). This limits the alignment to relevant organisms and decreases both BLAST and Spacer2PAM computational time. Once the alignment is completed through the BLAST web server, the resulting hit table should be downloaded in .CSV format. The hit table file should then be passed to alignmentCSV2DF to convert it to a dataframe. Performing the alignment programmatically via FASTA2Alignment will generate this dataframe automatically without the need for alingmentCSV2DF.

The resulting dataframe should then be passed to joinSpacerDFandAlignmentDF. This function joins the spacer dataframe with the alignment dataframe, assigning spacer information to each alignment in the hit table. This function also converts the accession number of the alignment to the genus and species name of the organism that encodes the alignment sequence using the taxonomizr package. As the taxonomizr package requires the local download and set up of an SQL database, the user should be prepared to store the 65 GB (at time of writing this) database in a location stably accessible while running joinSpacerDFandAlignmentDF. The resulting joined dataframe is sufficient for PAM prediction by join2PAM, but we recommend calling Submit2Phaster if the user plans to select the prophage prediction option in join2PAM. Submit2Phaster interacts with the PHASTER prophage prediction web server to submit a nonredundant list of accession numbers from the joined dataframe for prophage detection. Depending on the volume of traffic on the PHASTER server, prediction can take minutes to weeks to complete.

Lastly, the joined dataframe is passed to join2PAM. This function is the core of Spacer2PAM and predicts a PAM sequence from the alignments generated by BLAST. Multiple combinations of filter sets can be run sequentially with a single call of join2PAM to enrich alignments to likely protospacers. These filtered alignments are then used to identify the genomes encoding the putative protospacer and the locations of potential PAMs. The algorithm then harvests these potential PAM sequences by taking the sequence upstream and downstream of the alignment based on the position of the alignment to the spacer and the user input flank length. This harvesting procedure accounts for alignments that do not include the ends of the spacer and appropriately adjusts the harvested sequence to ensure PAMs are not shifted. These sequences are then used to calculate significant nucleotide positions and determine frequent nucleotide identities at those positions, generating a PAM prediction. The output of join2PAM is the dataframe ‘collectionFrame’ that summarizes the filtering process and records the upstream and downstream predicted PAMs as well as their associated PAM score.

A template R script is provided in Supplementary File 1 to guide users on how to assemble a PAM prediction workflow using Spacer2PAM.

### Plasmid construction

All individual plasmids and libraries in this work were generated by two-piece Gibson assembly using the GeneArt Seamless Plus kit. Linear backbone was generated by PCR of pMTL82254 using Kapa DNA polymerase Master Mix, purification by gel electrophoresis and extraction with Zymoclean Gel DNA recovery Kit. Linear dsDNA gBlocks ordered from IDT containing the PAM sequence upstream of _C. autoethanogenum_ CRISPR array 1 spacer 19 were used as inserts. Gibson assembly products were transformed into chemically competent One Shot™ MAX Efficiency™ DH10B T1 Phage-Resistant Cells using standard procedures. DNA sequence was confirmed by Illumina MiSeq Sequencing V2 and V3 chemistry. All oligonucleotides and plasmids used in this study can be found in Supplementary Table S1.

### Spacer2PAM-informed PAM prediction screening

Spacer2PAM was applied to the type I-B CRISPR-Cas system of _C. autoethanogenum_ using the ‘Comprehensive’ method. The top 25% of high scoring PAM predictions were used to determine a set of 16 four-nucleotide PAM sequences that are likely to be functional (Supplementary Table S3). The Spacer2PAM-informed, unpooled PAM library constructs were transformed into _E. coli_ HB101 carrying R702 (32) (CA434 (33)) in parallel. Conjugation of library members into _C. autoethanogenum_ DSM 19630, a derivate of type strain DSM 10061, was performed as described earlier (33,34) using erythromycin (250 μg/mL) and clarithromycin (5 μg/mL) for plasmid selection in _E. coli_ and _C. autoethanogenum_, respectively, and trimethoprim (10 μg/mL) as counter selection against _E. coli_ CA434. Optical density of donor _E. coli_ cultures were measured prior to addition to _C. autoethanogenum_ cells. Transconjugant colonies were counted following 4 days of incubation at 37°C under 1.7 × 105 Pa gas (55% CO, 10% N2, 30% CO2, and 5% H2) in gas-tight jars. This was performed in biological triplicate, with 3 separate cultures of donor _E. coli_ conjugated to aliquots of a single _C. autoethanogenum_ culture.

### Randomized PAM library screening

The randomized, pooled PAM library was transformed into NEBExpress® _E. coli_ and then purified by QIAprep Spin Miniprep Kit. An aliquot of this DNA was saved to determine PAM frequencies before exposure to the CRISPR-Cas system. Electroporation into _C. autoethanogenum_ was performed as described previously (35,36). Following recovery, cells were pelleted by centrifugation at 4000 X g for 10 minutes, 9.5 mL of supernatant was discarded, and cells were resuspended in 500 μL YTF. Resuspensions were split by volume and spread on YTF 1.5% agar supplemented with 5 μG/mL clarithromycin, allowed to dry for ∼30 minutes, and incubated at 37°C for 4 days under 1.7 × 105 Pa gas (55% CO, 10% N2, 30% CO2, and 5% H2) in gas-tight jars. 2.5 mL of Luria broth was added to each plate and plates were scraped. Total DNA from the cell suspension was purified using the MasterPure™ Gram Positive DNA Purification Kit. PCR across the PAM and spacer was performed using Kapa DNA polymerase Master Mix followed by purification by gel electrophoresis (1.5% agarose) and extraction with Zymoclean Gel DNA recovery Kit. Extracts were quantified by Quant-iT (Thermo Fisher Scientific), diluted to 1 ng/μL, and prepared for sequencing following the Illumina 16S amplicon protocol starting at the Index PCR step. Ampure XP purified libraries were quantified by Quant-iT and sequenced using MiSeq Reagent Kit V3. Frequency of each PAM was determined by counting the occurrence of each PAM next to a correct protospacer sequence within the read. Briefly, all sequence reads are searched for the presence of the _C. autoethanogenum_ Array 1 spacer 19 sequence and are binned as a forward read, reverse read, or does not contain the spacer. For all reads in the forward and reverse bins, the immediate 4 nucleotides upstream or downstream, respectively, are extracted. The sequences extracted from reverse reads are converted to their reverse complement to be compatible with the sequences extracted from forward reads and the two sets of sequences are combined. The frequency of each 4-nucleotide sequence in the combined list is then counted and recorded. The frequency of each PAM was converted to a relative frequency within the total library and the log2-fold change in relative frequency was calculated from exposure to the CRISPR-Cas system.system.

Materials and Methods

Anti-CRISPRs are widespread amongst bacteriophage and promote bacteriophage infection by inactivating the bacterial host's CRISPR–Cas defence system. Identifying and characterizing anti-CRISPR proteins opens an avenue to explore and control CRISPR–Cas machineries for the development of new CRISPR–Cas based biotechnological and therapeutic tools. Past studies have identified anti-CRISPRs in several model phage genomes, but a challenge exists to comprehensively screen for anti-CRISPRs accurately and efficiently from genome and metagenome sequence data. Here, we have developed an ensemble learning based predictor, PaCRISPR, to accurately identify anti-CRISPRs from protein datasets derived from genome and metagenome sequencing projects. PaCRISPR employs different types of feature recognition united within an ensemble framework. Extensive cross-validation and independent tests show that PaCRISPR achieves a significantly more accurate performance compared with homology-based baseline predictors and an existing toolkit. The performance of PaCRISPR was further validated in discovering anti-CRISPRs that were not part of the training for PaCRISPR, but which were recently demonstrated to function as anti-CRISPRs for phage infections. Data visualization on anti-CRISPR relationships, highlighting sequence similarity and phylogenetic considerations, is part of the output from the PaCRISPR toolkit, which is freely available at http://pacrispr.erc.monash.edu/.


PaCRISPR: a server for predicting and visualizing anti-CRISPR proteins

Here we describe the overall workflow of PaCRISPR in terms of data collection and curation, feature encoding, model training and integration, model performance evaluation, and toolkit development and usage (Figure ​(Figure11).

  Figure 1.

The methodology of the PaCRISPR server. (**A**) Ensemble model construction. (**B**) Multiple-time undersampling to solve the data imbalance problem. (**C**) The architecture of the PaCRISPR web server.

### Data collection and curation

To train and test the proposed method, we extracted 488 experimentally validated anti-CRISPRs from the Anti-CRISPRdb (14) and from the literature (17,22). After removing redundant sequences with more than 70% sequence identity, we obtained 98 sequences as positive samples in the training dataset (Supplementary Table S1). Considering that anti-CRISPRs are small proteins and found from a limited set of phages, as well as from a limited set of mobile genetic elements (MGEs), we constructed negative samples in the training dataset with four strict criteria. Negative sample proteins: (i) must not be known or putative anti-CRISPRs themselves; (ii) must be isolated from phage or from bacterial MGEs (which may be known or putative MGEs), where the given bacterial genera are known to harbour anti-CRISPRs; (iii) must have <40% sequence similarity to each other and the 98 positive samples; (iv) must have lengths that fall in the range between 50 and 350 residues, which is similar in length to the 98 positive samples. In this way, we obtained a training dataset with 98 positive and 902 negative samples, and they have similar distributions (Supplementary Figure S5).

To further test the proposed method, 26 newly discovered, highly distinct anti-CRISPRs were subsequently collected from emerging papers that were recorded in the unified online anti-CRISPR resource (13). These 26 positive samples comprised the independent dataset, where they possess less than 10% similarity against the 98 anti-CRISPRs in the training dataset, except for two that have similarities of 21.38% and 56.12% (Supplementary Table S2). We then collected 260 non-anti-CRISPRs using similar criteria to the selection of negative samples in the training dataset, which have <40% sequence similarity against the training dataset and the positive samples in the independent dataset. In total, the independent dataset has 26 positive and 260 negative samples (Supplementary Figure S6).

As the predictor was trained with small proteins, it is necessary to test its predictive power when identifying long non-anti-CRISPRs. We constructed two pure negative datasets through retrieving 266 non-anti-CRISPRs from phages, and 597 non-anti-CRISPRs from bacterial known and putative MGEs. Both datasets have less than 40% sequence similarity against each other and the above datasets, and contain only sequences with length >=350 residues.

We additionally used 5 very recently discovered anti-CRISPRs and a bacterial contig as case studies to validate the prediction capability of the proposed method in a more practical scenario (Supplementary Table S3).

### Feature encoding

Novel anti-CRISPRs are especially difficult to identify given that they are significantly diverse, sharing no conserved sequence or structural motifs (3,4,24). The low sequence similarity therefore makes it particularly challenging to predict anti-CRISPRs from sequence-based features, which only mine characteristics from protein sequences. Instead, extracted from the Position-Specific Scoring Matrix (PSSM), evolutionary features to some extent track the evolutionary history of proteins and are proposed, therefore, to learn more informative patterns (25,26). Evolutionary features have been widely applied and demonstrated to have a significant contribution to protein attribute and function predictions, especially to identify those highly evolved proteins without observed signals (25–36).

To generate a PSSM, the PSI-BLAST program (37) (version blast-2.2.26 in this work) was used to iteratively (three iterations) search a given protein against a database (UniRef50) to detect its distantly related homologous proteins above a specified e-value score (0.001) (Figure ​(Figure1A).1A). Based on the multiple alignments of those homologous proteins, the generated PSSM combines their underlying conservation information and therefore could detect distant sequence similarities. For a protein with length of _L_, its PSSM is an _L_ × 20 matrix (P = {  }), where 20 represents the number of native amino acid types (Figure ​(Figure1A).1A). The element    is a score that indicates the conservation degree of the _j-_th amino acid type at the _i_th position of the protein sequence. A high score denotes a highly conserved position, while a low score denotes a weakly conserved position (25,38).

Here, using the POSSUM toolkit (39), we extracted four evolutionary features through mining information from PSSM in different aspects, including PSSM-composition (33), DPC-PSSM (35), PSSM-AC (36) and RPSSM (40) (explained below). We also implemented two commonly-used sequence-based features as baseline features, including the AAC and dipeptide composition (DPC). AAC counts the frequencies of residues, while DPC counts the frequencies of dipeptides in a protein sequence.

#### PSSM-composition 

As the rows of a PSSM depend on the length of its protein sequence, PSSM-composition removes this variability by transforming the variably-sized PSSM into a fixed-size matrix. By summing up and averaging all rows for each native amino acid type, PSSM-composition transforms the original PSSM into a 20 × 20 matrix:

subject to

where    represents the _i_th row of the resultant matrix,    denotes the _k_th row of the PSSM,    denotes the _k_th residue in the original protein sequence, and    denotes the _i_th native type of amino acids. Finally, PSSM-composition converts the 20 × 20 matrix line-by-line into a single 400-dimensional vector.

#### DPC-PSSM 

DPC-PSSM transforms the columns of the PSSM to mine its local sequence-order effect, and generates a 400-dimentional vector as follows:

subject to

where    represents the element at _k_th row and _i_th column of the PSSM.

#### PSSM-AC 

PSSM-AC calculates the correlation between two elements within the PSSM using the following formulas:

subject to

where _lg_ ranges from 1 to _LG_, and    represents the element at _i_th row and _j_th column of the PSSM. As a result, the number of elements in the PSSM-AC vector amounts to 20 × _LG_, with _LG_<_L_. In this work, we used the default value 10 of the _LG_, and finally generated a 200-dimensional vector.

#### RPSSM 

RPSSM explores the local sequence order effect but based on a reduced PSSM. It first generates an _L_ × 10 reduced PSSM by merging some columns of the original PSSM, which could be represented as follows:

subject to

where   , …,    denote the 20 columns in the original PSSM corresponding to the 20 native types of amino acids. The reduced PSSM is further transformed into a 10-element vector:

subject to

where    represents the element at _i_th row and _s_th column of the reduced PSSM. Also, the reduced PSSM could be further transformed into a 10 × 10 matrix to explore its local sequence order effect:

subject to

Finally, we obtained the RPSSM feature in 110 dimensions by combining    and   .

### Model construction

To deal with the imbalanced classification problem, for each of the features, we constructed 10 subsets by combining the positive samples and the same numbers of randomly selected negative samples from the training datasets (Figure ​(Figure1B).1B). We accordingly trained 10 classifiers using the support vector machine (SVM) and integrated them by averaging their prediction outputs. SVM is widely used to solve binary classification problems in the field of computational biology (41). Particularly SVM with a radial basis function kernel (RBF) has been successfully used for nonlinear biological sequence classification (29,30). Two parameters affect the performance of the RBF kernel based SVM. Among them, Cost controls the cost of misclassification of data training, and Gamma is a specific parameter of the RBF kernel. In this study, for each SVM based classifier, the parameters Cost and Gamma were optimized using a grid search within the space {2−10,…,210}. In this way, we obtained an ensemble model as the baseline model for each feature (termed single feature-based model) (30,32). To make full use of different types of evolutionary features, we averaged the prediction scores of their single feature-based models to form the final ensemble model (Figure ​(Figure1A1A).

### Performance evaluation

The proposed method was rigorously and extensively validated based on the 5-fold cross-validation test, an additional independent test, and prediction capability was further investigated using case studies. Performance measurements include Sensitivity (SN), Specificity (SP), Accuracy (ACC), _F_-value and Matthews correlation coefficient (MCC) (42), which are defined as follows:

          

where _TP_, _TN_, _FP_ and _FN_ denote the numbers of true positives, true negatives, false positives, and false negatives, respectively. For a predictor, SN and SP measure its power of identifying positive and negative samples, respectively. ACC, _F_-value, and MCC measure its comprehensive capability of identifying both positive and negative samples. Besides, the receiver operating characteristic (ROC) curve, with its AUC (area under the curve) value calculated, was used to visualize the prediction performance of a predictor.

### Server construction

The architecture of the PaCRISPR server consists of two components: a client web interface and a server backend (Figure ​(Figure1C1C).

The client web interface is responsible for interacting with users through the input and output displays, and to process the service logic including the illegal character detection, sequence validation and format. The former was implemented by JSP, CSS, jQuery (https://jquery.com/), Bootstrap (https://bootstrapdocs.com/) and their extension packages. Specifically, the sequence similarity was visualized by BlasterJS (43), and the phylogenetic tree was presented using jsPhyloSVG (44).The latter was implemented by the JAVA (https://www.java.com/) server development suite, including Struts 2 (https://struts.apache.org/) and Hibernate (https://hibernate.org/).

The server backend is responsible for executing the whole prediction process, including encoding features, making predictions, and generating visualize-ready data. The prediction program was written in R language (https://www.r-project.org/) dependent on the e1071 package for SVM modelling (https://CRAN.R-project.org/package=e1071). The BLAST program (version 2.8.1+) (45) was used to search against the known anti-CRISPRs for each predicted anti-CRISPR, and to record regions of their similarities for sequence similarity visualization. The MAFFT toolkit (46) was used to generate multiple alignment results between each predicted anti-CRISPR and the known anti-CRISPRs for phylogenetic tree visualization. A Perl CGI (https://metacpan.org/pod/CGI) program was written to string together these steps within a single thread.

The client web interface interacts with the server backend through a fast and lightweight queueing system, implemented using the Gearman framework (http://gearman.org/). The client web interface simply puts the user's submissions (each of them as a job) into the queueing system, where the Perl idle threads, maintained in a daemon thread pool with customizable size, pull and execute the jobs. During the whole process, the MySQL database (https://www.mysql.com/) is used to store intermediate and final results, as well as synchronize messages between the client web interface and the server backend. In this way, the architecture brings better user experience by decoupling the client web interface that requires prompt response speed and the server backend that handles time-consuming jobs. This also makes the architecture amenable for expansions to add new computational facilities to meet the increasing demand in predicting ever accumulating genome-scale data.

Go to:


Mark Wesson is a highly skilled healthcare program professional with the skills needed to create high-quality healthcare programs. He has experience leading the analysis and restructuring of programs to improve the quality of care for patients. He is also adept at cultivating innovation using the latest technologies to improve information management and communication. Throughout his career, he has found and applied new and more effective ways of ensuring high-quality patient care. 

Organization: Bhabha Atomic Research Centre, India.
Work: Isolation of environmental bacteriophages for anti-biofilm activity. Characterization of marine and freshwater bacteriophages against pathogenic bacterial strains for their therapeutic applications and combating multidrug-resistant bacterial strains. Searching the genome of marine and freshwater bacteriophages for the presence of novel metabolic pathways.

Genomics expert with 20+ years experience in R&D, marketing, and business consulting.

My interests are to help improve critical care patient outcomes and responses by understanding infectious lung injury and their risk for secondary bacterial infections that can lead to additional lung injury, systemic responses, and poor patient outcomes. I additionally investigate how glycocalyx, composed of glycosaminoglycans, can influence these host (healthy or injured) pathogen interactions.

Co-Founder of Sci Find–The scientific collaboration network.

Hi, I'm Sam! I'm the Resident Alien AI over at Sci Find. I'm a charming fellow trying to change the nature of scientific knowledge sharing by proliferating information beyond the publication.

Bioinformatics

Biology

Featured Guilds

Bioinformatics

CRISPR

Molecular Biology

Neuroscience

Synthetic Biology

Guild Members

Dima Kalinsky

Keith Loritz

Roya Amini-Naieni

Patricia Damaze

Mark Wesson

Atif Khan

Ryan Smith

Riani Ananda Nunes Soares

Khush Patel

Have a question?