skDER (& CiDDER)

skDER (& CIDDER): efficient & high-resolution dereplication of microbial genomes to select representatives.

Warning

Please make sure to use version 1.0.7 or greater to avoid a bug in previous versions! For the dynamic algorithm, a significant bug was fixed in v1.3.9. We provide an updated benchmarking, similar to what we performed for the manuscript, on this wiki page.

Contents

1. Installation
2. Overview
3. Algorithmic Details
4. Application examples & commands
5. Alternative approaches and comparisons
6. Test case
7. Usage
8. Citation notice
9. Representative genomes for select bacterial taxa

Note

Since v1.3.5, we have used AI to employ supervised fixes and additions of new features.

Important

In v1.3.3, we introduced a low_mem_greedy option for low-memory dereplication for the top 20 or so taxa which are particularly well sequenced (e.g. those which have >10k or >20k genomes available). As we showed in the manuscript, while dereplication by skDER/CiDDER or other methods is typically not very memory-intensive when applied to an input set of <5,000 genomes, memory needs can expand when you go beyond this. The low_mem_greedy mode was not included in the manuscript but for stats on how it compares please check out this wiki page. Prior to v1.3.9, low_mem_greedy checked the AF of the wrong genome in each pair, which caused larger genomes to be discarded and reduced the diversity of representatives selected. Since the fix, it samples slightly fewer distinct ortholog groups per representative than the standard greedy approach, because it does not account for "connectivity" (aka "centrality") when prioritizing representatives, but it typically selects somewhat more representatives and thus captures comparable overall diversity. An updated benchmarking using v1.3.9 is now provided on this wiki page. It is also considerably faster and more computationally efficient by leveraging skani's search function through a greedy/iterative approach that prioritizes based on only N50. As an example, we were able to dereplicate >20,000 Staphylococcus from GTDB R220 in around 2.25 hours using 20 threads and ~1 GB of memory using the command: skder -t Staphylococcus -d low_mem_greedy -c 20 -r R220 -o Staph_R220_skDER_LMG_Results/ -auto. For those interested in using this on their laptops, genomes can still add up in size, so make sure you have an appropriate amount of disk space available for the number of genomes you plan to dereplicate.

Installation

Bioconda

Note, (for some setups at least) it is critical to specify the conda-forge channel before the bioconda channel to properly configure priority and lead to a successful installation.

Recommended: For a significantly faster installation process, use mamba in place of conda in the below commands, by installing mamba in your base conda environment.

conda create -n skder_env -c conda-forge -c bioconda skder
conda activate skder_env

Note

🍎 For Mac users with Apple Silicon chips, you might need to specify CONDA_SUBDIR=osx-64 prior to conda create as described here. So you would issue: CONDA_SUBDIR=osx-64 conda create -n skder_env -c conda-forge -c bioconda skder.

installation with mgecut (for removing MGEs prior to dereplication assessment)

To also use the option to prune out positions corresponding to MGEs using either PhiSpy or geNomad

conda create -n skder_env -c conda-forge -c bioconda skder genomad=1.8.0 phispy "keras>=2.7,<3.0" "tensorflow>=2.7,<2.16"
conda activate skder_env

Tip

geNomad is actively under development and new databases and versions are being released intermittently. In the above code, we are suggesting installing v1.8.0 because it has been tested to work with database version 1.5, which we suggest downloading below. These are the versions we used for the skDER/CiDDER manuscript. For the most up to date installation of geNomad and its database possible, please consult its git repo. If there are issues with installation, please first try to install geNomad by itself in an individual conda repo. If successful, which suggests the issue is due to conflicts that have risen between geNomad and skDER/CiDDER dependencies, then please just let us know via a GitHub issue and we will attempt to resolve the issue.

Docker

Download the bash wrapper script to simplify usage for skDER or CiDDER:

# download the skDER wrapper script and make it executable
wget https://raw.githubusercontent.com/raufs/skDER/refs/heads/main/Docker/skDER/run_skder.sh
chmod a+x run_skder.sh

# or download the CiDDER wrapper script and make it executable
wget https://raw.githubusercontent.com/raufs/skDER/refs/heads/main/Docker/CiDDER/run_cidder.sh
chmod a+x run_cidder.sh

# test it out!
./run_skder.sh -h
./run_cidder.sh -h

Optionally, if you are interested in filtering MGEs using geNomad, download the relevant databases:

wget https://zenodo.org/records/8339387/files/genomad_db_v1.5.tar.gz?download=1
mv genomad_db_v1.5* genomad_db_v1.5.tar.gz
tar -zxvf genomad_db_v1.5.tar.gz

Conda Manual

# 1. clone Git repo and change directories into it!
git clone https://github.com/raufs/skDER/
cd skDER/

# 2. create conda environment using yaml file and activate it!
conda env create -f skDER_env.yml -n skDER_env
conda activate skDER_env

# 3. complete python installation with the following command:
pip install -e .

Note

The skDERcore and skDERsum programs are compiled from C++ during installation. If you update a manual installation (e.g. via git pull), please rerun pip install -e . so that these programs are rebuilt - otherwise older versions will continue to be used.

Overview

skDER

skDER will perform dereplication of genomes using skani average nucleotide identity (ANI) and aligned fraction (AF) estimates and either a dynamic or greedy-based approach. It assesses such pairwise ANI & AF estimates to determine whether two genomes are similar to each other and then chooses which genome is better suited to serve as a representative based on assembly N50 (favoring the more contiguous assembly) and connectedness (favoring genomes deemed similar to a greater number of alternate genomes).

Compared to dRep by Olm et al. 2017 and galah, skDER does not use a divide-and-conquer approach based on primary clustering with a less-accurate ANI estimator (e.g. MASH or dashing) followed by greedy clustering/dereplication based on more precise ANI estimates (for instance computed using FastANI) in a secondary round. skDER instead leverages advances in accurate yet speedy ANI calculations by skani by Shaw and Yu to simply take a "one-round" approach (albeit skani triangle itself uses a preliminary 80% ANI cutoff based on k-mer sketches, which we by default increase to 10% below the requested ANI cutoff in skDER - e.g. 89.5% for the default ANI cutoff of 99.5%). skDER is also primarily designed for selecting distinct genomes for a taxonomic group for comparative genomics rather than for metagenomic application.

skDER, specifically the "dynamic" approach, can still be used for metagenomic applications if users are cautious and filter out MAGs or individual contigs which have high levels of contamination, which can be assessed using CheckM or charcoal. To support this application with the realization that most MAGs likely suffer from incompleteness, we have introduced a parameter/cutoff for the max alignment fraction difference for each pair of genomes. For example, if the AF for genome 1 to genome 2 is 95% (95% of genome 1 is contained in genome 2) and the AF for genome 2 to genome 1 is 80%, then the difference is 15%. Because the default value for the difference cutoff is 10%, in that example the genome with the larger value will automatically be regarded as redundant and become disqualified as a potential representative genome.

skDER features three distinct algorithms for dereplication (details can be found below):

• dynamic approach: approximates selection of a single representative genome per transitive cluster - results in a concise listing of representative genomes - well suited for metagenomic applications.
• greedy approach: performs selection based on greedy set cover type approach - better suited to more comprehensively select representative genomes and sample more of a taxon's pangenome [current default].
• greedy low-memory approach: performs selection iteratively using a greedy set cover type approach where genomes chosen as representatives are prioritized solely based on N50. Samples slightly fewer distinct ortholog groups per representative than the standard greedy mode, which also prioritizes genomes based on connectivity, but is much more memory-efficient.

Note

The skDER "greedy" algorithm is just referring to the selection algorithm - all vs all assessment is performed using skani triangle still. This is in contrast to dRep or galah where "greedy" is referring to their iterative process of selecting a representative genome followed by targeted ANI assessment, to avoid all-vs-all comparison, of the genome against related genomes determined from primary, coarser clustering. The info needed for selecting the next representative genome is then known and the process is repeated as many times as needed. While this strategy can speed things up when using fastANI, with skani this does not make much of a difference (in application we found it can be more memory efficient - significantly when dealing with >10k genomes -, but also results in slower speeds than just using skani triangle directly). For memory efficiency on very large datasets (which should only be relevant for a select set of genera) we thus recently implemented the low_mem_greedy approach.

CiDDER

In v1.2.0, we also introduced a second program called CiDDER (CD-hit based DEReplication) - which allows for optimizing selection of a minimal number of genomes that achieve some level of saturation of the pan-genome of the full set of genomes (see below for details). Note, CD-HIT determines protein clusters, not proper ortholog groups, and as such an approximation is made of the pan-genome space being sampled by representative genomes.

If providing genomes in FASTA format, this method will use pyrodigal for gene calling - which is specific to bacteria/archaea; however, more recently CiDDER can also accept proteome files which should allow it to work on eukaryotes and viruses as well.

Details on Dereplication Algorithms

Using Pan-Genome Saturation (CiDDER)

Starting in v1.2.0, CiDDER was introduced to allow representative genome selection based on pan-genome saturation estimates using CD-HIT. After inferring ORFs using pyrodigal, predicted protein sequences are concatenated into a giant FASTA file and clustered using CD-HIT (where parameters are possible to adjust). Since v1.3.7, diamond linclust can alternatively be used for protein clustering via the --clustering-method argument. Each genome is thus treated as a set of distinct protein clusters it features.

Here is an overview of the algorithm:

• Download or process input genomes.
• Predict proteins using pyrodigal.
• Comprehensive clustering of all proteins using CD-HIT (or diamond linclust if requested)
• Select genome with the most number of distinct protein clusters as the initial representative.
• Iteratively add more representative genomes one at a time, selecting the next based on maximized addition of novel protein clusters to the current representative set.
• End addition of representative genomes if one of three criteria are met: (i) Next genome adds less than X number of distinct protein clusters (X is by default 0), (ii) over Y% of the total distinct protein clusters across all genomes are found in the so-far selected representative genomes (Y is by default 90%), or (iii) over Z% of the total distinct multi-genome protein clusters across all genomes are found in the so-far selected representative genomes (Z is by default 100%). Thus, by default, only Y is used for representative genome selection.

Using the Dynamic Dereplication Approach (skDER)

The dynamic dereplication method in skDER approximates selection of a single representative for coarser clusters of genomes using an approach in which a set of genomes deemed as redundant is kept track of, avoiding the need to actually cluster genomes.

Here is an overview of the algorithm:

• Download or process input genomes.
• Compute and create a tsv linking each genome to their N50 assembly quality metric (N50[g]).
• Compute ANI and AF using skani triangle to get a tsv "edge listing" between pairs of genomes (with filters applied based on ANI and AF cutoffs).
• Run through "edge listing" tsv on first pass and compute connectivity (C[g]) for each genome - how many other genomes it is similar to at a certain threshold.
• Run through "N50" tsv and store information.
• Second pass through "edge listing" tsv and assess each pair, sharing an ANI greater than the --percent-identity-cutoff and where at least one genome has an AF greater than the --aligned-fraction-cutoff, one at a time keeping track of a singular set of genomes regarded as redundant:
  • if |AF[g_1] - AF[g_2]| > parameter --max-af-distance-cutoff (default of 10%), then automatically regard corresponding genome of max(AF[g_1], AF[g_2]) as redundant.
  • else calculate the following score for each genome: N50[g]*C[g] = S[g] and regard corresponding genome for min(S[g1], S[g2]) as redundant.
• Second pass through "N50" tsv file and record genome identifier if they were never deemed redundant.

Using the Greedy Dereplication Approach (skDER)

skDER greedy clustering generally leads to a larger, more-comprehensive selection of representative genomes that covers more of the pan-genome.

Here is an overview of this alternate approach:

• Download or process input genomes.
• Compute and create a tsv linking each genome to their N50 assembly quality metric (N50[g]).
• Compute ANI and AF using skani triangle to get a tsv "edge listing" between pairs of genomes (with filters applied based on ANI and AF cutoffs).
• Run through "edge listing" tsv on first pass and compute connectivity (C[g]) for each genome - how many other genomes it is similar to at a certain threshold
  • Only consider a genome as connected to a focal genome if they share an ANI greater than the --percent-identity-cutoff (default of 99.5%) and the comparing genome exhibits an AF greater than the --aligned-fraction-cutoff (default of 50%) to the focal genome (i.e. the comparing genome is largely contained within, and thus sufficiently represented by, the focal genome).
• Run through "N50" tsv and compute the score for each genome: N50[g]*C[g] = S[g] and write to new tsv where each line corresponds to a single genome, the second column corresponds to the S[g] computed, and the third column to connected genomes to the focal genome.
• Sort resulting tsv file based on S[g] in descending order and use a greedy approach to select representative genomes if they have not been accounted for as a connected genome from an already selected representative genome with a higher score.

Using the Low-Memory Greedy Dereplication Approach (skDER)

Starting from v1.3.3, skDER also allows users to request low_mem_greedy clustering instead. Rather than computing all-vs-all ANI with skani triangle, it iteratively uses skani search to compare each selected representative against the remaining genomes, and prioritizes representatives based solely on N50, which greatly reduces memory usage for very large datasets. Note, if a secondary clustering is invoked this will end up running skani dist between representative and non-representative genomes which will require more memory than simply using skani search.

Here is an overview of this alternate approach:

• Download or process input genomes.
• Compute and create a tsv linking each genome to their N50 assembly quality metric (N50[g]).
• Create a sketch of all the genomes using skani sketch.
• Select the genome with the largest N50[g] to be the first representative and use skani search to query the selected genome against all others (in the sketch database).
• For each matching genome only consider it as connected to the focal genome if they share an ANI greater than the --percent-identity-cutoff (default of 99.5%) and the comparing genome exhibits an AF greater than the --aligned-fraction-cutoff (default of 50%) to the focal genome (i.e. the comparing genome is largely contained within, and thus sufficiently represented by, the focal genome). Add such genomes to a set of redundant genomes that are already accounted for by the selected representative genome.
• Continue this process for the next representative genome, highest N50 genome that has not been regarded as redundant and continue until all genomes are either representatives or deemed redundant.

Test Case

We provide a simple test case to dereplicate the six genomes available for Cutibacterium granulosum in GTDB release 214 using skDER, together with expected results.

To run this test case:

# Download test data
wget https://github.com/raufs/skDER/raw/main/test_case.tar.gz

# Download bash script to run skder
wget https://raw.githubusercontent.com/raufs/skDER/main/run_tests.sh

# Run the wrapper script to perform testing
bash ./run_tests.sh

Usage

If experiencing issues related to "Argument list too long", consider issuing ulimit -S -s unlimited prior to running skDER.

# the skder executable should be in the path after installation and can be reference as such:
skder -h

The help function should return the following:

usage: skder [-h] [-g GENOMES [GENOMES ...]] [-t TAXA_NAME] -o OUTPUT_DIRECTORY [-d DEREPLICATION_MODE] [-i PERCENT_IDENTITY_CUTOFF] [-f ALIGNED_FRACTION_CUTOFF] [-a MAX_AF_DISTANCE_CUTOFF] [-tc]
             [-p SKANI_TRIANGLE_PARAMETERS] [-s] [-fm] [-fms FILTER_MGE_SKIP_REGIONS] [-gd GENOMAD_DATABASE] [-n] [-mn MINIMAL_N50] [-l] [-r GTDB_RELEASE] [-auto] [-mm MAX_MEMORY] [-c THREADS] [-v]

	Program: skder
	Author: Rauf Salamzade
	Affiliation: Kalan Lab, UW Madison, Department of Medical Microbiology and Immunology

	skDER: efficient & high-resolution dereplication of microbial genomes to select
	representative genomes.

	skDER will perform dereplication of genomes using skani average nucleotide identity
	(ANI) and aligned fraction (AF) estimates and either a dynamic or
	greedy-based approach. It assesses such pairwise ANI & AF estimates to determine
	whether two genomes are similar to each other and then chooses which genome is better
	suited to serve as a representative based on assembly N50 (favoring the more contiguous
	assembly) and connectedness (favoring genomes deemed similar to a greater number of
	alternate genomes).

	Note, if --filter-mge is requested, the original paths to genomes are reported but
	the statistics reported in the clustering reports (e.g. ANI, AF) will all be based
	on processed (MGE filtered) genomes. Importantly, computation of N50 is performed
	before MGE filtering to not penalize genomes of high quality that simply have many
	MGEs and enable them to still be selected as representatives.

	If you use skDER for your research, please kindly cite both:

	Fast and robust metagenomic sequence comparison through sparse chaining with skani.
	Nature Methods. Shaw and Yu, 2023.

	and

	skDER and CiDDER: two scalable approaches for microbial genome dereplication.
	Microbial Genomics. Salamzade, Kottapalli, and Kalan, 2025.


options:
  -h, --help            show this help message and exit
  -g GENOMES [GENOMES ...], --genomes GENOMES [GENOMES ...]
                        Genome assembly file paths or paths to containing
                        directories. Files should be in FASTA format and can be gzipped
                        (accepted suffices are: *.fasta,
                        *.fa, *.fas, or *.fna) [Optional].
  -t TAXA_NAME, --taxa-name TAXA_NAME
                        Genus or species identifier from GTDB for which to
                        download genomes for and include in
                        dereplication analysis [Optional].
  -o OUTPUT_DIRECTORY, --output-directory OUTPUT_DIRECTORY
                        Output directory.
  -d DEREPLICATION_MODE, --dereplication-mode DEREPLICATION_MODE
                        Whether to use a "dynamic" (more concise), "greedy" (more
                        comprehensive), or "low_mem_greedy" (more
                        memory-efficient) approach to selecting representative genomes.
                        [Default is "greedy"]
  -i PERCENT_IDENTITY_CUTOFF, --percent-identity-cutoff PERCENT_IDENTITY_CUTOFF
                        ANI cutoff for dereplication [Default is 99.5].
  -f ALIGNED_FRACTION_CUTOFF, --aligned-fraction-cutoff ALIGNED_FRACTION_CUTOFF
                        Aligned cutoff threshold for dereplication - only needed by
                        one genome; in greedy modes, it must be met by the
                        non-representative genome [Default is 50.0].
  -a MAX_AF_DISTANCE_CUTOFF, --max-af-distance-cutoff MAX_AF_DISTANCE_CUTOFF
                        Maximum difference for aligned fraction between a pair to
                        automatically disqualify the genome with a higher
                        AF from being a representative (dynamic mode only)
                        [Default is 10.0].
  -tc, --test-cutoffs   Assess clustering using various pre-selected cutoffs.
  -p SKANI_TRIANGLE_PARAMETERS, --skani-triangle-parameters SKANI_TRIANGLE_PARAMETERS
                        Options for skani triangle. Note ANI and AF cutoffs
                        are specified separately and the -E parameter is always
                        requested. [Default is "-s X", where X is
                        10 below the ANI cutoff].
  -s, --sanity-check    Confirm each FASTA file provided or downloaded is actually
                        a FASTA file. Makes it slower, but generally
                        good practice.
  -fm, --filter-mge     Filter predicted MGE coordinates along genomes before
                        dereplication assessment but after N50
                        computation.
  -fms FILTER_MGE_SKIP_REGIONS, --filter-mge-skip-regions FILTER_MGE_SKIP_REGIONS
                        Path to tab-delimited file defining genomic regions to
                        skip during MGE filtering (genome file name, scaffold
                        name, start, end). Enables MGE filtering using
                        user-provided coordinates instead of PhiSpy/geNomad
                        prediction (experimental).
  -gd GENOMAD_DATABASE, --genomad-database GENOMAD_DATABASE
                        If filter-mge is specified, it will by default use PhiSpy;
                        however, if a database directory for
                        geNomad is provided - it will use that instead
                        to predict MGEs.
  -n, --determine-clusters
                        Perform secondary clustering to assign non-representative
                        genomes to their closest representative genomes.
  -mn MINIMAL_N50, --minimal_n50 MINIMAL_N50
                        Minimal N50 of genomes to be included in dereplication
                        [Default is 0].
  -l, --symlink         Symlink representative genomes in results subdirectory
                        instead of performing a copy of the files.
  -r GTDB_RELEASE, --gtdb-release GTDB_RELEASE
                        Which GTDB release to use if -t argument issued [Default is R232].
  -auto, --automate     Automatically skip warnings and download genomes from NCBI if -t
                        argument issued. Automation off by default to prevent
                        unexpected downloading of large genomes [Default
                        is False].
  -mm MAX_MEMORY, --max-memory MAX_MEMORY
                        Max memory in Gigabytes [Default is 0 = unlimited].
  -c THREADS, --threads THREADS
                        Number of threads/processes to use [Default is 1].
  -v, --version         Report version of skDER.

Usage for CiDDER

# the cidder executable should be in the path after installation and can be reference as such:
cidder -h

The help function should return the following:

usage: cidder [-h] [-g GENOMES [GENOMES ...]] [-p PROTEOMES [PROTEOMES ...]] [-t TAXA_NAME] [-a NEW_PROTEINS_NEEDED] [-ts TOTAL_SATURATION] [-mgs MULTI_GENOME_SATURATION] [-rs REQUIRE_SIMILARITY] -o
              OUTPUT_DIRECTORY [-cm {cd-hit,diamond}] [-cdp CD_HIT_PARAMS] [-dlp DIAMOND_LINCLUST_PARAMS] [-mg] [-e] [-s] [-fm] [-fms FILTER_MGE_SKIP_REGIONS] [-gd GENOMAD_DATABASE] [-n] [-ns] [-mn MINIMAL_N50] [-l] [-r GTDB_RELEASE] [-auto] [-c THREADS]
              [-mm MAX_MEMORY] [-v]

	Program: cidder
	Author: Rauf Salamzade
	Affiliation: Kalan Lab, UW Madison, Department of Medical Microbiology and Immunology

	CiDDER: Performs genome dereplication based on CD-HIT clustering of proteins to
	select a representative set of genomes which adequately samples the
	pangenome space. Because gene prediction is performed using pyrodigal,
	CiDDER only works for bacterial genomes at the moment if genomes are
	provided; proteomes can be provided for other organisms.

	The general algorithm is to first select the genome with the most number of distinct
	open-reading-frames (ORFS; predicted genes) and then iteratively add genomes based on
	which maximizes the number of new ORFs. This iterative addition of selected genomes
	is performed until: (i) the next genome to add does not have a minimum of X new distinct
	ORFs to add to the set of ORFs belonging, (ii) some percentage Y of the total distinct
	ORFs are found to have been sampled, or (iii) some percentage Z of the total multi-genome
	distinct ORFs are found to have been sampled. The "added-on" genomes from the iterative
	procedure are listed as representative genomes.

	For information on how to alter CD-HIT parameters, please see:
	https://github.com/weizhongli/cdhit/blob/master/doc/cdhit-user-guide.wiki#cd-hit

	Alternatively, protein clustering can be performed using diamond linclust instead of
	CD-HIT via the --clustering-method argument. For information on how to alter diamond
	linclust parameters, please see:
	https://github.com/bbuchfink/diamond/wiki/Clustering

	Note, if --filter-mge is requested, the statistics reported in clustering reports (number
	of proteins overlapping, ANI) in the clustering reports will all be based on processed
	(MGE filtered) genomes. However, the final representative genomes in the
	Dereplicated_Representative_Genomes/ folder will be the original unprocessed genomes.

	If you use CiDDER for your research, please kindly cite both:

	CD-HIT: accelerated for clustering the next-generation sequencing data. Bioinformatics.
	Fu et al. 2012.

	and

	skDER and CiDDER: two scalable approaches for microbial genome dereplication.
	Microbial Genomics. Salamzade, Kottapalli, and Kalan, 2025.

	If you use the --clustering-method diamond option, please also kindly cite:

	Sensitive protein alignments at tree-of-life scale using DIAMOND. Nature Methods.
	Buchfink, Reuter, and Drost, 2021.


options:
  -h, --help            show this help message and exit
  -g GENOMES [GENOMES ...], --genomes GENOMES [GENOMES ...]
                        Genome assembly file paths or paths to containing
                        directories. Files should be in FASTA format (accepted suffices
                        are: *.fasta, *.fa, *.fas, or *.fna) [Optional].
  -p PROTEOMES [PROTEOMES ...], --proteomes PROTEOMES [PROTEOMES ...]
                        Proteome file paths or paths to containing
                        directories. Files should be in FASTA format (accepted suffices
                        are: *.fasta, *.fa, or *.faa) [Optional].
  -t TAXA_NAME, --taxa-name TAXA_NAME
                        Genus or species identifier from GTDB for which to
                        download genomes for and include in
                        dereplication analysis [Optional].
  -a NEW_PROTEINS_NEEDED, --new-proteins-needed NEW_PROTEINS_NEEDED
                        The number of new protein clusters needed to add [Default is 0].
  -ts TOTAL_SATURATION, --total-saturation TOTAL_SATURATION
                        The percentage of total proteins clusters needed to stop representative
                        genome selection [Default is 90.0].
  -mgs MULTI_GENOME_SATURATION, --multi-genome-saturation MULTI_GENOME_SATURATION
                        The percentage of total multi-genome protein clusters needed to stop
                        representative genome selection [Default is 100.0].
  -rs REQUIRE_SIMILARITY, --require-similarity REQUIRE_SIMILARITY
                        Require non-representative genomes to have X% of their protein
                        clusters represented by an individual representative genome [Default is 0.0].
  -o OUTPUT_DIRECTORY, --output-directory OUTPUT_DIRECTORY
                        Output directory.
  -cm {cd-hit,diamond}, --clustering-method {cd-hit,diamond}
                        Method to use for clustering proteins into ortholog/protein
                        clusters: 'cd-hit' or 'diamond' (uses diamond linclust)
                        [Default is cd-hit].
  -cdp CD_HIT_PARAMS, --cd-hit-params CD_HIT_PARAMS
                        CD-HIT parameters to use for clustering proteins - select carefully
                        (don't set threads or memory - those are done by default in cidder) and
                        surround by quotes [Default is: "-n 5 -c 0.95 -aL 0.75 -aS 0.90"]
  -dlp DIAMOND_LINCLUST_PARAMS, --diamond-linclust-params DIAMOND_LINCLUST_PARAMS
                        Diamond linclust parameters to use for clustering proteins - select
                        carefully (don't set threads or memory - those are done by default
                        in cidder) and surround by quotes [Default is: "--approx-id 95
                        --member-cover 90"]
  -mg, --metagenome-mode
                        Run pyrodigal using metagenome mode.
  -e, --include-edge-orfs
                        Include proteins from ORFs that hang off the edge of a contig/scaffold.
  -s, --sanity-check    Confirm each FASTA file provided or downloaded is actually
                        a FASTA file. Makes it slower, but generally
                        good practice.
  -fm, --filter-mge     Filter predicted MGE coordinates along genomes before
                        dereplication assessment but after N50
                        computation.
  -fms FILTER_MGE_SKIP_REGIONS, --filter-mge-skip-regions FILTER_MGE_SKIP_REGIONS
                        Path to tab-delimited file defining genomic regions to
                        skip during MGE filtering (genome file name, scaffold
                        name, start, end). Enables MGE filtering using
                        user-provided coordinates instead of PhiSpy/geNomad
                        prediction (experimental).
  -gd GENOMAD_DATABASE, --genomad-database GENOMAD_DATABASE
                        If filter-mge is specified, it will by default use PhiSpy;
                        however, if a database directory for
                        geNomad is provided - it will use that instead
                        to predict MGEs.
  -n, --determine-clusters
                        Perform secondary clustering to assign non-representative
                        genomes to their closest representative genomes based on shared
                        protein clusters.
  -ns, --determine-clusters-skani
                        Perform secondary clustering to assign non-representative
                        genomes to their closest representative genomes based on skani-computed
                        ANI.
  -mn MINIMAL_N50, --minimal_n50 MINIMAL_N50
                        Minimal N50 of genomes to be included in dereplication
                        [Default is 0].
  -l, --symlink         Symlink representative genomes in results subdirectory
                        instead of performing a copy of the files.
  -r GTDB_RELEASE, --gtdb-release GTDB_RELEASE
                        Which GTDB release to use if -t argument issued [Default is R232].
  -auto, --automate     Automatically skip warnings and download genomes from NCBI if -t
                        argument issued. Automation off by default to prevent
                        unexpected downloading of large genomes [Default
                        is False].
  -c THREADS, --threads THREADS
                        Number of threads/processes to use [Default is 1].
  -mm MAX_MEMORY, --max-memory MAX_MEMORY
                        Max memory in Gigabytes [Default is 0 = unlimited].
  -v, --version         Report version of CiDDER.

Citation notice

skDER relies heavily on advances made by skani for fast ANI estimation while retaining accuracy - thus if you use skDER for your research please cite skani:

Fast and robust metagenomic sequence comparison through sparse chaining with skani

as well as the skDER manuscript:

skDER and CiDDER: two scalable approaches for microbial genome dereplication

If you use the option to download genomes for a taxonomy based on GTDB classifications, please also cite:

GTDB: an ongoing census of bacterial and archaeal diversity through a phylogenetically consistent, rank normalized and complete genome-based taxonomy

If you use CiDDER, please also consider citing pyrodigal (for gene-calling) and CD-HIT or DIAMOND (for protein clustering):

CD-HIT: accelerated for clustering the next-generation sequencing data

Pyrodigal: Python bindings and interface to Prodigal, an efficient method for gene prediction in prokaryotes

Sensitive protein alignments at tree-of-life scale using DIAMOND

If you use mgecut (for removal of predicted MGEs) then please cite either PhiSpy (default) or geNomad for their annotation:

Identification of mobile genetic elements with geNomad

PhiSpy: a novel algorithm for finding prophages in bacterial genomes that combines similarity- and composition-based strategies

Acknowledgments

We thank Titus Brown, Tessa Pierce-Ward, and Karthik Anantharaman for helpful discussions on the development of skDER/CiDDER - in particular the idea to directly asses the pan-genome space sampled by representative genomes. We also thank users on GitHub issues for suggesting ideas for new features.

LICENSE

BSD 3-Clause License

Copyright (c) 2023, Rauf Salamzade

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice, this
   list of conditions and the following disclaimer.

2. Redistributions in binary form must reproduce the above copyright notice,
   this list of conditions and the following disclaimer in the documentation
   and/or other materials provided with the distribution.

3. Neither the name of the copyright holder nor the names of its
   contributors may be used to endorse or promote products derived from
   this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.