local-cd-search 0.3.0

Protein annotation using local PSSM databases from CDD

local-cd-search

A command-line tool for local protein domain annotation using NCBI's Conserved Domain Database (CDD).

Background

NCBI CD-Search is a widely used tool for functional annotation of proteins. It uses RPS-BLAST to search protein sequences against position-specific scoring matrices (PSSMs) from the CDD database. PSSMs offer higher sensitivity for detecting distant homologs than searches against individual protein sequences, while remaining substantially faster than HMM-based annotation.

While the CD-Search web interface is convenient for small queries, it is not well suited for large-scale annotation. local-cd-search enables local protein annotation and automates the entire workflow: downloading PSSM databases from CDD, running RPS-BLAST, post-processing results with rpsbproc to filter hits using CDD's curated bit-score thresholds.

Installation

The easiest way to install local-cd-search is with Pixi, which will manage dependencies automatically and make local-cd-search available for execution from anywhere.

pixi global install -c conda-forge -c bioconda local-cd-search

Alternatively, you can install it from PyPI. In this case, rpsblast and rpsbproc must be installed separately. To install local-cd-search from PyPI using uv, run:

uv tool install local-cd-search

Quick start

Download PSSM databases

Download the full CDD database, which is a collection of six individual databases (see table below):

local-cd-search download database cdd

Or download individual databases. For example:

# COG database
local-cd-search download database cog

# Multiple databases
local-cd-search download database cog pfam tigr

The databases available for download are:

Database	Name	Description
cdd	CDD	Collection of PSSMs derived from multiple sources (all databases listed below except KOG)
cdd_ncbi	NCBI-curated domains	Domain models that leverage 3D structural data to define precise boundaries
cog	COG	Groups of orthologous Prokaryotic proteins
kog	KOG	Groups of orthologous Eukaryotic proteins
pfam	Pfam	Large collection of protein families and domains from diverse taxa
prk	PRK	NCBI collection of protein clusters containing reference sequences from prokaryotic genomes
smart	SMART	Models of domains from proteins involved in signaling, extracellular, and regulatory functions
tigr	TIGRFAM	Manually curated models for functional annotation of microbial proteins

Annotate proteins

To run annotation on a FASTA file of protein sequences (in this example, proteins.faa) and save results to results.tsv, run the following command:

local-cd-search annotate proteins.faa results.tsv database

The local-cd-search will automatically detect which databases are available and will them for annotation.

Output

The output of local-cd-search annotate is a tab-separated file with hits filtered by CDD's curated bit-score thresholds. The following columns are included:

Column	Description
query	Protein identifier
hit_type	Specific, Non-specific, or Superfamily
pssm_id	CDD PSSM identifier
from	Start position in query
to	End position in query
evalue	E-value
bitscore	Bit score
accession	Domain accession
short_name	Domain short name (e.g., COG0001)
incomplete	Indicates if there are more than 20% missing from the N- or C- terminal ends (-, N, C, or NC)
superfamily_pssm_id	Superfamily PSSM identifier

hit_type

• Specific: The top-ranking RPS-BLAST hit (compared to other hits in overlapping intervals) that meets or exceeds a domain-specific E-value threshold. It represents a very high confidence that the query sequence belongs to the same protein family as the sequences used to create the domain model.
• Non-specific: Hits that meet or exceed the RPS-BLAST threshold for statistical significance (default E-value cutoff of 0.01).
• Superfamily: The domain cluster to which the specific and/or non-specific hits belong. This is a set of conserved domain models that generate overlapping annotation on the same protein sequences and are assumed to represent evolutionarily related domains.

incomplete

Indicates if there are more than 20% missing from the N- or C- terminal compared to the original domain. Possible values are:

• -: No more than 20% shorter on either terminals.
• N: N-terminal has 20% or more missing.
• C: C-terminal has 20% or more missing.
• NC: Both terminals have 20% or more missing.

Functional sites output

If --sites-output is specified, an additional tab-separated file is created with functional site annotations. The following columns are included:

Column	Description
query	Protein identifier
annot_type	Specific or Generic
title	Description of the functional site
coordinates	Residues and their positions (e.g., H94,Y96)
complete_size	Total number of residues in the site
mapped_size	Number of residues mapped to the query
source_domain	PSSM ID of the domain where the site is defined

Usage

download subcommand

local-cd-search download [OPTIONS] DB_DIR DATABASE...

Option	Short	Argument	Description	Default
--force		flag	Force re-download even if files are already present.
--quiet		flag	Suppress non-error console output.
--help	-h	flag	Show help message and exit.

annotate subcommand

local-cd-search annotate [OPTIONS] INPUT_FILE OUTPUT_FILE DB_DIR

Option	Short	Argument	Description	Default
--evalue	-e	FLOAT (≥ 0)	Maximum allowed E-value for hits.	0.01
--ns		flag	Include non-specific hits in the output results table.
--sf		flag	Include superfamily hits in the output results table.
--threads		INTEGER	Number of threads to use for rpsblast.	0
--sites-output	-s	FILE	Path to write functional site annotations.
--data-mode	-m	std | rep | full	Redundancy level of domain hit data passed to rpsbproc: rep (best model per region of the query), std (best model per source per region), full (all models meeting E-value significance).	std
--tmp-dir		DIRECTORY	Directory to store intermediate files. If not specified, temporary files will be deleted after execution.
--quiet		flag	Suppress non-error console output.
--help	-h	flag	Show help message and exit.