DomainMapper

DomainMapper is an HMM parser designed to annotate proteins' domain structures from sequence alone using the ECOD database.
DomainMapper annotates non-contiguous, insertional, and circularly permuted domains as well.
DomainMapper works on Linux / MacOS / Windows

Dependencies

In order to use DomainMapper you will need numpy, scipy and bio These can be installed using pip

Installing Dependencies

# installing numpy
pip install numpy

# installing scipy
pip install scipy

# installing BioPython
pip install bio

Installing DomainMapper

cd /DomainMapper
pip install .

# or

pip install ./DomainMapper

Running DomainMapper

dommap -f raw_hmmscan_output.hmm.out -o mapped_protein_domains.mapped.out

Documentation

usage: dommap [-h] [-f F] [-o O] [--ecod_domains ECOD_DOMAINS] [--intra_gap INTRA_GAP] [--inter_gap INTER_GAP] [--overlap OVERLAP] [--frac_overlap FRAC_OVERLAP] [--eval_cutoff EVAL_CUTOFF] [--update]

arguments:
  -h, --help            show this help message and exit
  -f F                  Input path to file from 'hmmscan'
  -o O                  Output path for mapped domains
  --dom_def DOM_DEF     Path to ECOD 'Latest Domains' text file (default = file is automatically downloaded [165 MB Free Space Required (deleted
                        after parsing)] [2 MB File Saved])
  --intra_gap INTRA_GAP, --intra_domain_gap_tolerance INTRA_GAP
                        Optional minimum gap size within a high-scoring pair for those residues to be carved out, generating a non-contiguous hit
                        (default = 30)
  --inter_gap INTER_GAP, --inter_domain_gap_tolerance INTER_GAP
                        Optional minimum gap size between two high-scoring pairs for the residues inbetween to be left out, generating a non-
                        contiguous hit (default = 30)
  --overlap OVERLAP, --domain_overlap_tolerance OVERLAP
                        Optional overlap between high-scoring pairs to mandate an elimination (default = 40)
  --frac_overlap FRAC_OVERLAP, --fractional_domain_overlap_tolerance FRAC_OVERLAP
                        Optional fractional overlap between high-scoring pairs to mandate an elimination (0.0 - 1.0) (default = 0.7)
  --eval_cutoff EVAL_CUTOFF
                        Optional upper bound tolerance of the E-value (default = 1e-5)
  --update              Update ECOD 'Latest Domains'

HMMR3 & DomainMapper Tutorial

To begin using DomainMapper, clone the repository to a destination on your local computer

0. git clone https://github.com/FriedLabJHU/DomainMapper

To use DomainMapper, the full output from an HMMR3 hmmscan is required as an input.
Here is how to generate this file:

1. Get a file containing protein sequences in .fasta format (for instance, a proteome file from uniprot)

2. Obtain the ECOD HMM profile database

3. Place this file on your computer, untar and unzip tar -xzvf ecodf.hmm.tar.gz

At this point, you will need to run hmmscan. The user needs to have separately installed HMMER3.

Installing HMMR3

# On MacOS (Intel Only)
brew install hmmer

# On Ubuntu (x86/x64 Processors Only)
apt install hmmer

Currently there are no versions of HMMER for Windows but HMMER3 on WSL could be a good alternative

Additionally, there are currently no fully-supported distributions of HMMER on ARM-bases processors

Visit http://hmmer.org/documentation.html for details.

4. hmmpress ecodf.hmm

5. hmmscan -o your_hmmscan_output.hmm.out /path/to/ecodf.hmm /path/to/your_fasta_file.fasta

6. dommap -f /path/to/your_hmmscan_output.hmm.out -o /path/to/your_hmmscan_output.mapped.out

To read the various additional options available, read the help docstring
dommap -h

Reading DomainMapper Output

The first 31 lines consist of a header containing input and output file names, and parameter values. This is followed by summary statistics for the domains identified in the FASTA file submitted.

Specifically:

• Total Domains - Self explanatory
• NC - Number of non-contiguous domains
• IS - Number of insertional domains
• CP - Number of circularly permutated domains

The second part of the output contains the list of identified domains. Each of which is assigned values in 9 columns (for ECOD searches).

• Column 1 - Protein indentifier from FASTA
• Column 2 - Conditional E-value
• Column 3 - Domain Range
• Column 4 - Domain Property (NC, IS, CP)
• Column 5 - ECOD Architecture (Will be NA if ECOD HMMs are not used)
• Column 6 - ECOD X-Group (Will be NA if ECOD HMMs are not used)
• Column 7 - ECOD T-Group (Will be NA if ECOD HMMs is not used)
• Column 8 - ECOD F-Group
• Column 9 - ECOD F-ID (Will be NA if ECOD HMMs is not used)

Domains with complex topologie (i.e. nesting, weaving, etc.) can be identified by the "NC IS" property flags in column 4.

Citation

Manriquez-Sandoval, E, Fried, SD. DomainMapper: Accurate domain structure annotation including those with non-contiguous topologies. Protein Science. 2022; 31( 11):e4465. https://doi.org/10.1002/pro.4465

Funding

• NSF Division of Molecular and Cellular Biology (2045844)

• NIH Training Grant - Program in Molecular Biophysics (T32GM135131)