Pipeline imitates the workflow in a scientific paper (Danso et.al.) which generates a Hidden Markov Model from known 9 PETases and searchs databases with this Hidden Markov Model to discover novel PETases.

In this snakemake pipeline we built a HMM with same 9 molecules and search only 1 metagenomic dataset(not in the repo check Dataset section).

This search results in 2 proteins under our threshold constraints(hmmsearch score > 150). Then we generate a phylogenetic tree with those 2 and 9 known PETases.

• python (tested with 3.10.0)
• snakemake (tested with 6.12.1)
• conda
• Other dependencies that will be automatically downloaded by snakemake:
  • tcoffee
  • hmmer
  • fasttree

There are two types of data in this pipeline.

• Known PETases
• Metagenomic datasets

Known PETases are automatically downloaded by pipeline by their Uniprot entry keys identified in config/config.json. Metagenomic datasets in this study cannot be downloaded by APIs. Thus users are excepted to download those metagenomic datasets manually and move to data/metagenomic_datasets/ folder and fill the necessary parts in config/config.json. The metagenomic dataset used in this study is here. Moving 2140918008.a.faa file into data/metagenomic_datasets/ and filling config/config.json would be sufficient to reproduce results.

Pipeline generates its final output, which is a phylogenetic tree in Newick format, in data/pyhylogenetic_tree_PETases.tree. This tree can be visualized by online tools. We used this tool. Proteins with Bog_all are from the metagenomic dataset.