Sphae

Phage toolkit to detect phage candidates for phage therapy

Overview

The steps that sphae takes are shown here:

This snakemake workflow was built using Snaketool [https://doi.org/10.1371/journal.pcbi.1010705], to assemble and annotate phage sequences. Currently, this tool is being developed for phage genomes. The steps include,

• Quality control that removes adaptor sequences, low-quality reads and host contamination (optional).
• Assembly
• Contig quality checks; read coverage, viral or not, completeness, and assembly graph components.
• Phage genome annotation

Cite Sphae: https://doi.org/10.1093/bioadv/vbaf004

If you are new to bioinformatics or running command line tools, here is a great tutorial to follow: https://github.com/AnitaTarasenko/sphae/wiki/Sphae-tutorial

Install

Pip install

```bash

creating a new envrionment

conda create -y -n sphae python=3.12 conda activate sphae

install sphae

pip install sphae ```

Conda install

Setting up a new conda environment

bash conda create -n sphae python conda activate sphae

Container Install There are two versions of the container

1. Sphae v1.5.2 Includes databases, so the container is about 32GB

Steps to donwload and run this container

``` TMPDIR= IMAGEDIR-

singularity pull --tmpdir $TMPDIR --dir $IMAGEDIR docker://npbhavya/sphae:latest
singularity exec sphae_latest.sif sphae --help
singularity exec sphae_latest.sif sphae run --help


singularity exec -B <path/to/inputfiles>:/input,<path/to/output>:/output sphae_latest.sif sphae run --input /input --output /output

```

1. Sphae v1.5.2-noDB This version, doesnt come with databases

Here are the commands to download sphae container ``` TMPDIR= IMAGEDIR-

singularity pull --tmpdir $TMPDIR --dir $IMAGEDIR docker://npbhavya/sphae:v1.5.2-noDB
singularity exec sphae_v1.5.2-noDB.sif sphae --help
singularity exec sphae_v1.5.2-noDB.sif sphae run --help

# <path/to/databases> set to sphae/workflow/databases if sphae install is run 
singularity exec -B <path/to/databases>:/database,<path/to/inputfiles>:/input,<path/to/output>:/output sphae_latest.sif sphae run --input /input --output /output
```

Source install

```bash

clone sphae repository

git clone https://github.com/linsalrob/sphae.git

move to sphae folder

cd sphae

install sphae

pip install -e .

confirm the workflow is installed by running the below command

sphae --help ```

Installing databases

Run the below command,

```bash

Installs the database to default directory, sphae/workflow/databases

sphae install

Install database to specific directory

sphae install --db_dir ```

Install the databases to a directory, sphae/workflow/databases

This workflow requires the - Pfam35.0 database to run viral_verify for contig classification. - CheckV database to test for phage completeness - Pharokka databases - Phynteny models - Phold databases - Medaka models

This step requires ~23G of storage If these databases are already installed, skip this step and instead set the envrionment variables pointing to the where these databases are installed

```bash

Note to change the file path to the databases.

For instance if sphae was installed using conda, the databases by default will be saved to /home/username/miniforge3/envs/sphae/lib/python3.11/site-packages/sphae/workflow/databases

export VVDB=sphae/workflow/databases/Pfam35.0/Pfam-A.hmm.gz export CHECKVDB=sphae/workflow/databases/checkv-db-v1.5 export PHAROKKADB=sphae/workflow/databases/pharokka_db export PHYNTENYDB=sphae/workflow/databases/models export PHOLDDB=sphae/workflow/databases/phold ```

Running the workflow

Sphae is developed to be modular: - sphae run will run QC, assembly and annotation - sphae annotate will run only annotation steps

Commands to run

Only one command needs to be submitted to run all the above steps: QC, assembly and assembly stats

```bash

For illumina reads, place the reads both forward and reverse reads to one directory

Make sure the fastq reads are saved as {samplename}R1.fastq and {samplename}R2.fastq or with extensions {samplename}R1.fastq.gz

sphae run --input tests/data/illumina-subset --output example -k

For nanopore reads, place the reads, one file per sample in a directory

sphae run --input tests/data/nanopore-subset --sequencing longread --output example -k

For newer ONT sequencing data where polishing is not required, run the command

sphae run --input tests/data/nanopore-subset --sequencing longread --output example -k --no_medaka

To run either of the commands on the cluster, add --executor slurm to the command. There is a little bit of setup to do here.

Setup a ~/.config/snakemake/slurm/config.yaml file - https://snakemake.github.io/snakemake-plugin-catalog/plugins/executor/slurm.html#advanced-resource-specifications

I may have set this workflow to run only slurm right now, will make it more generic soon.

sphae run --input tests/data/nanopore-subset --preprocess longread --output example --profile slurm -k --threads 16 ```

Command to run only annotation steps and phylogenetic trees This step reruns - Pharokka, Phold, Phynteny - Phylogenetic tree with terminase large subunit, portal protein

```bash

the genomes directory has the already assembled complete genomes

run the export commands to set the database paths

sphae annotate --genome --output example -k ```

Output

Output is saved to example/RESULTS directory. In this directory, there will be four files - Genome annotations in GenBank format (Phynteny output) - Genome in fasta format (either the reoriented to terminase output from Pharokka, or assembled viral contigs) - Circular visualization in png format (Pharokka output) - Genome summary file

Genome summary file includes the following information to help, - Sample name - Length of the genome - Coding density - If the assembled contig is circular or not (From checkv) - Completeness (calculated from CheckV) - Contamination (calculated from CheckV) - Taxonomy accession ID (Pharokka output, searches the genome against INPHARED database using mash) - Taxa mash includes the number of matching hashes of the assembled genome to the accession ID/Taxa name. Higher the matching hash- more likely the genome is related to the taxa predicted - Gene searches: - Whether integrase is found (search for integrase gene in annotations) - Whether anti-microbial genes were found (Phold and Pharokka search against AMR database) - Whether any virulence factors were found (Pharokka search against virulence gene database) - Whether any CRISPR spacers were found (Pharokka search against MinCED database)

FAQ

1. "Failed during assembly":

   • This message indicates that the assembly process was unsuccessful. It suggests that the assembler could not generate contigs, which are contiguous sequences of DNA, typically representing segments of a genome.
   • To confirm this, you can check the logs located at sphae.out/PROCESSING/assembly/flye/<sample name>/assembly_info.txt or sphae.out/PROCESSING/assembly/megahit/<sample name>/log. These logs should provide details about the error or the step at which the assembly failed.
   • One possible reason for this failure could be insufficient genome coverage, meaning that there were not enough sequencing reads to accurately assemble the genome.

2. "Genome includes multiple contigs, fragmented":

   • This message indicates that the assembly generated numerous short fragments (contigs) instead of a single, contiguous sequence representing a nearly complete phage genome.
   • You can verify this by examining the file sphae.out/PROCESSING/assembly/flye/<sample name>-assembly-stats_flye.csv or sphae.out/PROCESSING/assembly/megahit/<sample name>-assembly-stats_megahit.csv.
   • Each row in these tables represents a contig along with its characteristics. If none of the contigs are identified as viral and do not meet a certain completeness threshold (e.g., greater than 70% completeness), it suggests that the assembly consists of fragmented contigs.
   • Fragmented contigs make it challenging to accurately identify genes. To address this issue, you may need to resequence the phages for better coverage or try using different assembly algorithms.

3. "Good genome coverage but still encountering assembly issues":

   • If you have adequate genome coverage but still face assembly problems, you may consider adjusting the subsampling step in sphae. This step involves randomly selecting a subset of reads to reduce the computational burden.
   • To modify the subsampling parameters, navigate to the config/config.yaml file and update the line under subsample section, for example: subsample: --bases 1000M
   • Increase or decrease the number of bases (e.g., 1000M for 1000 megabases) based on your requirements.
   • After making the changes, rerun sphae and ensure that the updated subsampling parameters are reflected in the sphae.out/sphae.config.yaml file.

4. "What does 'No integrases found ...but Phynteny predicted a few unknown function genes to have some similarity with integrase genes but with low confidence. Maybe a false positive or a novel integrase gene' mean?" This message indicates that while no integrase genes were explicitly identified, the analysis detected certain genes that exhibited similarities to integrase genes. However, these genes were associated with low confidence scores, suggesting a possibility of being false positives or potentially representing novel integrase genes.

Phynteny, the tool used for this prediction, assigns a confidence score to each gene prediction. If this score falls below a certain threshold (typically 90%), the gene remains classified as having an unknown function. To further investigate these genes, advanced techniques such as folding using tools like ColabFold and Foldseek can be employed. Analyzing the structure of these genes may provide additional insights into their functionality and potential role in biological processes.

1. How do I visualize the phages and gene annotations? To visualize the phages and gene annotations, I recommend using Clinker. First, gather all the sample genbank files from sphae.out/RESULTS and place them in a new directory. Then, execute the clinker command to generate clinker plots, which compare the genes in each genome to each other.

Additionally, for enhanced visualization, consider running dnaapler on the genomes in fasta format obtained from sphae.out/RESULTS. This step generates reoriented phages that start with terminase genes. Pharokka -> Phold -> Phynteny has to be rerun, and the resulting genbank files can be used for visualization. To perform the annotation steps, run the command sphae annotate --input <reoriented genomes from dnaapler in fasta format directory>

Please note that dnaapler may fail if terminase genes are not found, particularly when working with novel phages. The reason these steps haven't been added to sphae. If you encounter any challenges during this process, please feel free to leave an issue, and I'll provide improved documentation to assist you further with the command on how to install and run the command different commands.

1. Where are the intermediate files being saved? These files are being saved in sphae.out/PROCESSING. If you need more information on the file structure here, or have ideas of better organization then leave an issue and I will make a note to have more documentation.

2. Just run annotation on already assembled genomes?

   sphae annotate --input <input genomes> This command runs only Pharokka, Phold and Phynteny to annotate the assembled genomes. The results are saved to a new directory labeled sphae.out/annotation.

   Note: Currently, Sphae runs Phold in CPU mode, but efforts are underway to support Phold GPU mode for faster processing of this step.

3. How to change the number of base pairs to subsample for a sample? Run the command sphae config This copies the config file within the workflow to the current directory. Open this file and update the line bases: 10000000 to for instance bases: 300000 Then run sphae run with the command sphae run --input tests/data/illumina-subset --output example -k --config <path to the config file with the change>

Citation

To cite sphae, doi: https://doi.org/10.1101/2024.11.18.624194

Issues and Questions

If you come across any issues or errors, report them under Issues.