MHG stands for Maximal Homologous Group. Inputting genome nucleotide seqeunces, MHG is an annotation-free graph-based tool to merge and partition homologous groups, and outputs homologous groups for the target genome set where each group has its evolutionary history as a single tree and involves no rearrangements. The below sections introduce three sub-programs and you can find an testing example case in the latter section:
- MHG: Start from nucleotide sequences to build blastn databases and queries. And then partition for MHGs.
- genome-to-blast-db: Start from nucleotide sequences, only build blastn databases and queries.
- MHG-partition: Start from Blastn queries, only partition for MHGs.
Depending on your input, you can choose the corresponding executables. MHG takes whole genome sequences and do all the jobs including building BLASTn databases and obtaining queries, which integrates both genome-to-blast-db and MHG-partition, and outputs MHGs into a txt file where each line is a MHG.
It is highly recommended to setup a new conda environment to avoid weird stuck through conda install! Installing MHG via conda will save the time figuring out the dependencies.
conda create --name mhg python=3.7
conda activate mhg
conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge
conda install -c bioconda mhg
If you encounter errors running directly conda install -c bioconda mhg, try the three conda config above;
If you are stuck on "solving environment", run conda config --remove channels conda-forge, and then conda config --add channels conda-forge should solve the problem.
But again, it is highly recommend to create a brand new environment.
Using git clone, please install the below dependencies manually:
Also, there are required built-in python packages:
Please make sure to add the below two executables(genome-to-blast-db, MHG-partition) as two environment variables when using MHG if you installed via git clone instead of conda.
usage: MHG [-h] [-g GENOME] [-b BLAST] [-db DATABASE] [-q QUERY] [-w WORD_SIZE] [-T THREAD] [-go GAPOPEN] [-ge GAPEXTEND] [-o OUTPUT] [-t THRESHOLD]
Make blastn database & Build blastn queries
optional arguments:
-h, --help show this help message and exit
-g GENOME, --genome GENOME
Genome nucleotide sequence directory (Required)
-b BLAST, --blast BLAST
Blastn bin directory; try to call 'makeblastdb, blastn' straightly if no path is inputted by default(if blast folder is added as an environemnt variable
-db DATABASE, --database DATABASE
Directory to store blast nucleotide databases for each sequence in genome directory. By default write to current folder 'blastn_db'
-q QUERY, --query QUERY
Output folder storing all blastn queries in xml format. By defualt write to current folder 'blastn_against_bank'
-w WORD_SIZE, --word_size WORD_SIZE
Blastn word size, default 28
-T THREAD, --thread THREAD
Blastn thread number, default 1
-go GAPOPEN, --gapopen GAPOPEN
Blastn gap open penalty, default 5
-ge GAPEXTEND, --gapextend GAPEXTEND
Blastn gap extend penalty, default 2
-o OUTPUT, --output OUTPUT
File containing the final partitioned MHGs, each line represents a MHG containing different blocks
-t THRESHOLD, --threshold THRESHOLD
Bitscore threshold for determining true homology
usage: genome-to-blast-db [-h] [-g GENOME] [-b BLAST] [-db DATABASE] [-q QUERY] [-w WORD_SIZE] [-T THREAD] [-go GAPOPEN] [-ge GAPEXTEND]
Make blastn database & Build blastn queries
optional arguments:
-h, --help show this help message and exit
-g GENOME, --genome GENOME
Genome nucleotide sequence directory (Required)
-b BLAST, --blast BLAST
Blastn bin directory; try to call 'makeblastdb, blastn' straightly if no path is inputted by default(if blast folder is added as an environemnt variable
-db DATABASE, --database DATABASE
Directory to store blast nucleotide databases for each sequence in genome directory. By default write to current folder 'blastn_db'
-q QUERY, --query QUERY
Output folder storing all blastn queries in xml format. By defualt write to current folder 'blastn_against_bank'
-w WORD_SIZE, --word_size WORD_SIZE
Blastn word size, default 28
-T THREAD, --thread THREAD
Blastn thread number, default 1
-go GAPOPEN, --gapopen GAPOPEN
Blastn gap open penalty, default 5
-ge GAPEXTEND, --gapextend GAPEXTEND
Blastn gap extend penalty, default 2
usage: MHG-partition [-h] [-q QUERY] [-o OUTPUT] [-t THRESHOLD]
Partition and generate modules
optional arguments:
-h, --help show this help message and exit
-q QUERY, --query QUERY
Input folder for module partition storing all blastn queries in xml format
-o OUTPUT, --output OUTPUT
File containing the final partitioned modules, each line represents a module containing different blocks
-t THRESHOLD, --threshold THRESHOLD
Bitscore threshold for determining true homology
MHG -g genomes/ -t 0.95
The four test bacteria are stored in the folder genomes, you can git clone to get it!
The command builds blastn databases and runs blastn queries, and partition for MHGs for four sample bacteria locatated in genomes with a bitscore threshold 0.95 (A Query with bit score above 95% of maximum bit score is considered as a true homology; a more detailed description of the threshold can be found in our paper). This example should run about 20-30 minute outputted to mhg.txt where each line is a MHG. This should have about 1000 MHGs.
MHG outputs a txt file where each line is an individual MHG. For each MHG, it contains at least two blocks, and each block is in a format:
((sequence_accession,(union_start,union_end)),(homology_start,homology_end),direction)
sequence_accession is the contig ID from a genome, for example Escherichia coli O104 has a RefSeq assembly accession GCF_000299455.1, and it contains four contigs: chromosome NC_018658.1, plasmid NC_018666.1, plasmid NC_018659.1, plasmid NC_018660.1. sequence_accession will be starting from "NC_" in this case.
union_start,union_end are for internal purpose while traversing the alignment graph and they can be ignored while parsing the MHG results.
homology_start,homology_end denote the start and end nucleotide indices in the sequence_accession, which indicates the actual homologous subsequence position.
direction is a relative direction to the other blocks from the same line MHG, it is a value either "+" or "-" indicating forward or reverse complement respectively.