SamnTrek: Locally deployable search tool to explore local Salmonella epidemiology in a global context

SamnTrek enables rapid integration of NCBI Pathogen Detection (NPD) data to place local Salmonella strains in a global context. Specifically, given any Salmonella genomes, SamnTrek identifies closely matching strains in NPD and constructs a phylogeny to contextualize the (local) query sequences with close matching (global) hits. The sequence similarity search algorithm has been optimized extensively, rendering it possible to search through >400,000 Salmonella genomes in minutes using minimal resources.

Quick start

This section is dedicated to those who cannot be bothered with the fine details and are simply seeking to get the pipeline up and running asap.

Minimal prerequisites

• One of the supported containerization platforms installed: Singularity or Docker
• Nextflow >= v23.0.0
• Git
• Internet connection

Once all the prerequisites have been installed, run the following command in the terminal to verify all dependencies have been set up correctly:

bash nextflow run jimmyliu1326/SamnTrek -r [version] --help

[!NOTE] Replace [version] with the latest release tag, which can be found here

If successful, the complete pipeline help message will be printed to screen.

Configuring the input samplesheet

The input sequence data must be formatted as a .csv file containing two columns: sample and genome

• The sample column contains unique identifiers to each query genome i.e. sample ID
• The genome column contains the path to each query genome
• The sample sheet must contain column headers sample and genome

Example samplesheet:

sample,genome Sample_A,/path/to/Sample_A.fasta Sample_B,/path/to/Sample_B.fasta

Downloading the database

Precomputed NPD database can be downloaded by running the following command:

nextflow run jimmyliu1326/SamnTrek \ -r [version] \ -profile <docker/singularity> \ --wf download_db \ --download_path /path/to/SamnTrek_db

• -profile indicates which containerization platform to use for process execution i.e. Docker or Singularity
• --wf download_db signals the pipeline to only download the database
• --download_path specifies where to save the downloaded database files

With all of the above steps completed, the full pipeline can be executed by specifying --wf all

nextflow run jimmyliu1326/SamnTrek \ -r [version] \ -profile <docker/singularity> \ --wf all \ --input samplesheet.csv \ --outdir ./results --db /path/to/SamnTrek_db

• --input specifies the path to input samplesheet
• --outdir specifies the path to save the output results
• --db specifies the path to the directory containing predownloaded database files

Modular Pipeline Design

SamnTrek orchestrates the logic flow between four primary modules that collectively enable rapid integration of NPD Salmonella data. The modular design allows flexible points of entry. Users can resume from any steps in the pipeline without restarting the pipeline from scratch, which can expedite parameter tuning, testing and optimization.

The key modules of SamnTrek include (listed in chronological order of operation): 1. Sort (sort) - Placement of query in precomputed genomic clusters 2. Search (search) - Search against NPD sequences belonging to the same cluster and identify the subset of closest matching hits using unsupervised clustering 3. Fetch (fetch_hits) - Downloaded the full genomes of the close matches from NCBI 4. Contextualize (contextualize) - Construct a phylogeny integrating both the query and close matching hits

For example, users can quickly produce close matching hits based on different search parameters without reperforming the prior sort step.

[!NOTE] While Nextflow does have a built-in resume function, it relies on the integrity of the temporary files stored in working directory. However, in some institutions, these working directories are routinely cleansed to free up storage space. Hence, we have implemented a custom method to reuse cached results directly from the output directory (--outdir)

Results in a previous output directory can be reused by supplying the directory path using the --results_dir option

```bash

Initial run using top_hits = 100

nextflow run jimmyliu1326/SamnTrek \ --wf all \ --input samplesheet.csv \ --outdir ./results \ --db ./SamnTrekdb \ --tophits 100

Second run using top_hits = 200 and

reusing previous results

nextflow run jimmyliu1326/SamnTrek \ --wf search,fetchhits,contextualize \ --input samplesheet.csv \ --resultsdir ./results \ --outdir ./newresults \ --db ./SamnTrekdb \ --top_hits 200 ```

Output files legend

The following table describes the content stored in each subdirectory within the output directory (--outdir).

| File | Description | | :-- | :-- | | SORTRESULTS/ | Cluster placement results organized by sample ID. | SORTRESULTS/.tsv | Describes the predicted cluster assignments made by three different methods (best hit search, KNN, phylogenetic placement) and the final assignment based on majority voting. | SEARCHRESULTS/ | Sequence similarity search results organized by sample ID. | SEARCHRESULTS/.hits | Contains the NCBI accession IDs of the close matching strains. | SEARCHRESULTS/*.tsv | Contains the estimated core and accessory distance to all the database sequences belonging to the same cluster as the query. | SEARCHRESULTS/searchstats.tsv | Summarizes the subject to (top) hits ratio | | SEARCHRESULTS/coreaccessoryplot.png | Scatterplot displaying the distribution of relative distances of subject sequences to the query | | SEARCHRESULTS/hdbscanclusterscore.png | Results of HDBSCAN hyperparameter tuning evaluating cluster quality score (silhouette index) and total cluster count | | PDGENOMES/ | Archives all close matching genomes downloaded from NCBI | | SUMMARY/ | Summary files compiling results across all samples supplied in the input sample sheet | | SUMMARY/sortresults.tsv | Aggregated cluster placement predictions | | SUMMARY/searchresults.tsv | Aggregated sequence search results | | SUMMARY/SamnTreksummary.tsv | Complete summary file aggregating cluster placement and sequence search results | | TREE/ | Data related to phylogenetic analysis | | TREE/.nwk | Raw phylogenetic tree file | | TREE/.distancematrix.tsv | Pairwise distance matrix used in phylogenetic construction |

Future plans

• Real-time synchronization with NPD
• Support core genome SNV phylogenetics
• Streamline tree visualizations by facilitating the compilation of available metadata of the close matching hits from NPD

Credits

SamnTrek was originally written by Jimmy Liu.

Citations

This pipeline uses code and infrastructure developed and maintained by the nf-core community, reused here under the MIT license.

SamnTrek manuscript is currently in progress. It will be updated here when available.

The nf-core framework for community-curated bioinformatics pipelines.

Philip Ewels, Alexander Peltzer, Sven Fillinger, Harshil Patel, Johannes Alneberg, Andreas Wilm, Maxime Ulysse Garcia, Paolo Di Tommaso & Sven Nahnsen.

Nat Biotechnol. 2020 Feb 13. doi: 10.1038/s41587-020-0439-x.