linda5mith / crassify

Pipeline to taxonomically classify viral metagenome data.

crassify

Pipeline to taxonomically classify microbiome viruses in metagenomic data based on ICTV's VMR.

1. Generate matches.tsv using diamond v2 [https://github.com/bbuchfink/diamond]

Diamond input: your translated, unknown viral proteins

Diamond blastp your reference proteins against ICTV reference DB

For symmetrical best hit approach use --query-cover 50 --subject-cover 50 parameters else run with as recommended below:

diamond blastp --query proteins.faa --db databases/ictv_reference.dmnd -o matches.tsv --ultra-sensitive --no-self-hits --evalue 0.00001

2. Installing crassify

Install dependencies using environment.yml file: conda env create -f environment.yml

conda activate crassify

git clone https://github.com/linda5mith/crassify.git

Set path to crassify installation PATH=$PATH:/home/user/programs/crassify

3. Running crassify

Sample folders with test inputs are provided in /proteomes and /diamond_hits/matches.tsv but will be replaced with the path to your proteomes and the matches.tsv file generated from step 1.

python crassify.py -p /proteomes -m /diamond_hits/matches.tsv -db /metadata/ICTV_metadata.csv

4. Crassify output

protein_hits.csv - returns most homologous ICTV viral protein hit for each of your viral proteins that aligned to reference database.

genome_hits.csv - returns top 5 genome hits for each proteome.

dists.csv - returns distances between all proteomes and related ICTV viral genomes.