Taxonomy-specific-RPS-BLAST-part-I

Protein is usually made up of domain(s). Conserved Domains can be described by local multiple sequence alignments spanning a variety of organisms to reveal sequence regions that contain the same, or similar, patterns of amino acids. Although it is easy to retrieve the taxonomical distribution of a protein, it is not available at the domain level.

This 2019 NCBI-Codeathon project will develop a pipeline to assign a lowest common taxid to a conserved protein domain (defined by a Position-Specific Score Matrix, PSSM). The taxid represents the taxon that contains this domain specifically with given threshold. This project will be incorporated into the future RPS-Blast (RPS-BLAST uses the query sequence to search a database of pre-calculated PSSMs, and report significant hits) to provide taxonomic information in CD Search (Conserved Domain Search) results.

Dependencies:

Python 3

BioPython

NCBI Conserved Domain Architecture Retrieval Tool (CDART)

taxidlineage.dmp extracted from new_taxdump.zip available in NCBI Taxonomy ftp site (ftp://ftp.ncbi.nlm.nih.gov/pub/taxonomy/new_taxdump/)

Workflow:

Input:

A table generate by sql query against NCBI internal CDART database (example file)

Command:

./dtrt.py <data/*.tsv-file-name> [-threshold <number-between-0-and-1> ]

options:

• -threshold: Threshold to report taxonomy node. Default value: 0.95

• -show_names: Display taxonomic names instead of just taxids. Default value: false

• -show_tree: Display taxonomy tree for model. Default value: false

• -shake: Experimental: 'shakes' the tree to remove nodes that contribute less than 1% to the parent's weight. Default value: false

Output:

A taxonomy tree with the lowest common taxid with the threshold for the domain

Validation:

PSSM-Id: 129695,200311,334026,334050,337780,335786,274086,308214,315456,338615,287328,313550,313551,274263

Future Work

• Elaborate weight by considering taxonomic origin (e.g.: metagenomic, plasmids, synthetic/artificial) may count for less
• Use model-specific thresholds to get better matches
• Elaborate weight by providing higher weight to those sequences that have higher evalue/bitscore
• Instead of IPG, go to the individual sequences and extract the taxid
• From Ryan Connor: weights to consider the variation that sequences bring into the model

People/Team

Marc Gwadz, IEB/NCBI/NIH

Christiam Camacho, IEB/NCBI/NIH

Jianli Dai, IEB/NCBI/NIH

Hanguan Liu, IEB/NCBI/NIH

Mingzhang Yang, IEB/NCBI/NIH