16S SNAPP is an analysis workflow to be run on Linux/Mac command line interface (CLI) for 16S multi-V region amplicon sequences from Swift's 16S SNAP panel. It generates taxonomic composition tables at and above genus ranks from multiple demultiplexed fastq files.

16S SNAPP's approach is to associate sequence reads derived from multiple amplicon regions to their most probable sequences of origin, i.e. the assumed templates. This is done through database search (blastn) for high identity matches among reference sequences followd by intersecting aligned reads on the matching references, read count allocation for multi-mapped reads, and classification of consensus sequences. It offers higher sensitivity for 16S multiple-amplicon data compared to tools designed for single 16S amplicon data.

Setup:

1. Install and setup the following software: R (https://www.r-project.org/), Java ≥1.8.0_131, Python 3.6.8 or later (with Numpy ≥ 1.16.0, Pandas ≥ 1.1.4, Scipy ≥1.4.1), DADA2 (https://benjjneb.github.io/dada2/), BLAST (ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/), RDPTools (github.com/rdpstaff/RDPTools), Cutadapt (https://cutadapt.readthedocs.io/en/stable/), VSEARCH (https://github.com/torognes/vsearch), MAFFT (https://mafft.cbrc.jp/alignment/software/, FASTTREE (http://www.microbesonline.org/fasttree/) minimap2 2.20-r1061 (https://github.com/lh3/minimap2)
2. Create a folder , e.g. 'DB', and download to it the reference and primer files from (https://danaherlifesciences.box.com/s/fkojslg8t0xnksvyd6vwp2gsfdacmi5p).
3. Clone this repository (git clone https://github.com/swiftbiosciences/16S-SNAPP-py3)
4. Edit “config.txt” to enter absolute paths to tools, 'DB' and primer file, and the expected single read length after primer is trimmed

Command to run: snapp.sh config.txt inputdir workdir

Input: gzipped post-demultiplexing sequence fastq files each (pair) representing a single sample and each carrying a Swift 16S primers on its 5’ end.

Output: (“RESDIR” folder under the directory where the command is run):

      lineage-table.tsv,
      feature-table.tsv,
      taxonomy-table.tsv,
      templates_mafft.tree (experimental)
      OTU_count.txt (experimental)
      OTU_taxonomy.txt (experimental)

Limitations:

1. The workflow largely relies on aligning reads from different gene regions to their closest reference sequences from the database (currently packed with RDP11.5). Nevertheless, sequence pairs that can’t be aligned to any reference sequences within the set identity cutoff are treated as individual features and classified directly.
2. The resolution of taxonomic assignments depends on the classifier, mainly its taxonomic coverage of the samples, and the sequencing depth. The current version of 16S SNAPP uses RDP Classifier, but is potentially compatible with other k-mer's based classifier.
3. The phylogenetic tree is built from the sequence alignment of the assumed template sequences to approximate the phylogeny of multiple amplicons that inherently lack comparable alignment positions.
4. OTU table ('OTU_count.txt' paired with 'OTU_taxonomy.txt' file) is generated by combinig unmapped ASVs and consensus sequences clustered after pairwise alignments using minimap2. It is an attempt to consolidate features represented by consensus sequences for better resolutions in sample comparisons.

Additional notes: Quality trimming/filtering/denoising may be adjusted in run_dada2.R script as needed for specific data sets to improve the performance of the analysis.