# GetOrganelle 1.7.5.0 ## GetOrganelle This toolkit assemblies organelle genome from genomic skimming data. It achieved the best performance overall both on simulated and real data and was recommended as the default for chloroplast genome assembly in a third-party comparison paper ([Freudenthal et al. 2020. Genome Biology](https://doi.org/10.1186/s13059-020-02153-6)). ## Conda env activation You must activate the conda env to use this software. To activate: ```console bash source /local/cluster/getorganelle/activate.sh ``` Location and version: ```console $ which get_organelle_from_reads.py /local/cluster/getorganelle/bin/get_organelle_from_reads.py $ get_organelle_from_reads.py --version GetOrganelle v1.7.5 $ echo $GETORG_PATH /nfs1/CGRB/databases/GetOrganelle/latest ``` help message: ```console $ get_organelle_from_reads.py --help usage: ### Embryophyta plant plastome, 2*(1G raw data, 150 bp) reads get_organelle_from_reads.py -1 sample_1.fq -2 sample_2.fq -s cp_seed.fasta -o plastome_output -R 15 -k21,45,65,85,105 -F embplant_pt ### Embryophyta plant mitogenome get_organelle_from_reads.py -1 sample_1.fq -2 sample_2.fq -s mt_seed.fasta -o mitogenome_output -R 30 -k 21,45,65,85,105 -F embplant_mt GetOrganelle v1.7.5 get_organelle_from_reads.py assembles organelle genomes from genome skimming data. Find updates in https://github.com/Kinggerm/GetOrganelle and see README.md for more information. optional arguments: -v, --version show program's version number and exit -h print brief introduction for frequently-used options. --help print verbose introduction for all options. IN-OUT OPTIONS: Options on inputs and outputs -1 FQ_FILE_1 Input file with forward paired-end reads (format: fastq/fastq.gz/fastq.tar.gz). -2 FQ_FILE_2 Input file with reverse paired-end reads (format: fastq/fastq.gz/fastq.tar.gz). -u UNPAIRED_FQ_FILES Input file(s) with unpaired (single-end) reads (format: fastq/fastq.gz/fastq.tar.gz). files could be comma-separated lists such as 'seq1.fq,seq2.fq'. -o OUTPUT_BASE Output directory. Overwriting files if directory exists. -s SEED_FILE Seed sequence(s). Input fasta format file as initial seed. A seed sequence in GetOrganelle is only used for identifying initial organelle reads. The assembly process is purely de novo. Should be a list of files split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). Default: '/nfs1/CGRB/databases/GetOrganelle/latest/SeedDatabase/*.fasta' (* depends on the value followed with flag '-F') -a ANTI_SEED Anti-seed(s). Not suggested unless what you really know what you are doing. Input fasta format file as anti-seed, where the extension process stop. Typically serves as excluding plastid reads when extending mitochondrial reads, or the other way around. You should be cautious about using this option, because if the anti-seed includes some word in the target but not in the seed, the result would have gaps. For example, use the embplant_mt and embplant_pt from the same plant-species as seed and anti-seed. --max-reads MAXIMUM_N_READS Hard bound for maximum number of reads to be used per file. A input larger than 536870911 will be treated as infinity (INF). Default: 1.5E7 (-F embplant_pt/embplant_nr/fungus_mt/fungus_nr); 7.5E7 (-F embplant_mt/other_pt/anonym); 3E8 (-F animal_mt) --reduce-reads-for-coverage REDUCE_READS_FOR_COV Soft bound for maximum number of reads to be used according to target-hitting base coverage. If the estimated target-hitting base coverage is too high and over this VALUE, GetOrganelle automatically reduce the number of reads to generate a final assembly with base coverage close to this VALUE. This design could greatly save computational resources in many situations. A mean base coverage over 500 is extremely sufficient for most cases. This VALUE must be larger than 10. Set this VALUE to inf to disable reducing. Default: 500. --max-ignore-percent MAXIMUM_IGNORE_PERCENT The maximum percent of bases to be ignore in extension, due to low quality. Default: 0.01 --phred-offset PHRED_OFFSET Phred offset for spades-hammer. Default: GetOrganelle-autodetect --min-quality-score MIN_QUALITY_SCORE Minimum quality score in extension. This value would be automatically decreased to prevent ignoring too much raw data (see --max-ignore- percent).Default: 1 ('"' in Phred+33; 'A' in Phred+64/Solexa+64) --prefix PREFIX Add extra prefix to resulting files under the output directory. --out-per-round Enable output per round. Choose to save memory but cost more time per round. --zip-files Choose to compress fq/sam files using gzip. --keep-temp Choose to keep the running temp/index files. --config-dir GET_ORGANELLE_PATH The directory where the configuration file and default databases were placed. The default value also can be changed by adding 'export GETORG_PATH=your_favor' to the shell script (e.g. ~/.bash_profile or ~/.bashrc) Default: /nfs1/CGRB/databases/GetOrganelle/latest SCHEME OPTIONS: Options on running schemes. -F ORGANELLE_TYPE This flag should be followed with embplant_pt (embryophyta plant plastome), other_pt (non-embryophyta plant plastome), embplant_mt (plant mitogenome), embplant_nr (plant nuclear ribosomal RNA), animal_mt (animal mitogenome), fungus_mt (fungus mitogenome), fungus_nr (fungus nuclear ribosomal RNA)or embplant_mt,other_pt,fungus_mt (the combination of any of above organelle genomes split by comma(s), which might be computationally more intensive than separate runs), or anonym (uncertain organelle genome type). The anonym should be used with customized seed and label databases ('-s' and '--genes'). For easy usage and compatibility of old versions, following redirection would be automatically fulfilled without warning: plant_cp->embplant_pt; plant_pt->embplant_pt; plant_mt->embplant_mt; plant_nr->embplant_nr --fast ="-R 10 -t 4 -J 5 -M 7 --max-n-words 3E7 --larger-auto-ws --disentangle-time- limit 360" This option is suggested for homogeneously and highly covered data (very fine data). You can overwrite the value of a specific option listed above by adding that option along with the "--fast" flag. You could try GetOrganelle with this option for a list of samples and run a second time without this option for the rest with incomplete results. --memory-save ="--out-per-round -P 0 --remove-duplicates 0" You can overwrite the value of a specific option listed above by adding that option along with the "--memory- save" flag. A larger '-R' value is suggested when "--memory-save" is chosen. --memory-unlimited ="-P 1E7 --index-in-memory --remove-duplicates 2E8 --min-quality-score -5 --max-ignore-percent 0" You can overwrite the value of a specific option listed above by adding that option along with the "--memory-unlimited" flag. EXTENDING OPTIONS: Options on the performance of extending process -w WORD_SIZE Word size (W) for pre-grouping (if not assigned by '--pre-w') and extending process. This script would try to guess (auto-estimate) a proper W using an empirical function based on average read length, reads quality, target genome coverage, and other variables that might influence the extending process. You could assign the ratio (1>input>0) of W to read_length, based on which this script would estimate the W for you; or assign an absolute W value (read length>input>=35). Default: auto-estimated. --pre-w PREGROUP_WORD_SIZE Word size (W) for pre-grouping. Used to reproduce result when word size is a certain value during pregrouping process and later changed during reads extending process. Similar to word size. Default: the same to word size. -R MAX_ROUNDS, --max-rounds MAX_ROUNDS Maximum number of extending rounds (suggested: >=2). Default: 15 (-F embplant_pt), 30 (-F embplant_mt/other_pt), 10 (-F embplant_nr/animal_mt/fungus_mt/fungus_nr), inf (-P 0). --max-n-words MAXIMUM_N_WORDS Maximum number of words to be used in total.Default: 4E8 (-F embplant_pt), 2E8 (-F embplant_nr/fungus_mt/fungus_nr/animal_mt), 2E9 (-F embplant_mt/other_pt) -J JUMP_STEP The length of step for checking words in reads during extending process (integer >= 1). When you have reads of high quality, the larger the number is, the faster the extension will be, the more risk of missing reads in low coverage area. Choose 1 to choose the slowest but safest extension strategy. Default: 3 -M MESH_SIZE (Beta parameter) The length of step for building words from seeds during extending process (integer >= 1). When you have reads of high quality, the larger the number is, the faster the extension will be, the more risk of missing reads in low coverage area. Another usage of this mesh size is to choose a larger mesh size coupled with a smaller word size, which makes smaller word size feasible when memory is limited.Choose 1 to choose the slowest but safest extension strategy. Default: 2 --bowtie2-options BOWTIE2_OPTIONS Bowtie2 options, such as '--ma 3 --mp 5,2 --very-fast -t'. Default: --very- fast -t. --larger-auto-ws By using this flag, the empirical function for estimating W would tend to produce a relative larger W, which would speed up the matching in extending, reduce the memory cost in extending, but increase the risk of broken final graph. Suggested when the data is good with high and homogenous coverage. --target-genome-size TARGET_GENOME_SIZE Hypothetical value(s) of target genome size. This is only used for estimating word size when no '-w word_size' is given. Should be a list of INTEGER numbers split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). Default: 130000 (-F embplant_pt) or 390000 (-F embplant_mt) or 13000 (-F embplant_nr) or 39000 (-F other_pt) or 13000 (-F animal_mt) or 65000 (-F fungus_mt) or 13000 (-F fungus_nr) or 39000,390000,65000 (-F other_pt,embplant_mt,fungus_mt) --max-extending-len MAX_EXTENDING_LEN Maximum extending length(s) derived from the seed(s). A single value could be a non-negative number, or inf (infinite) or auto (automatic estimation). This is designed for properly stopping the extending from getting too long and saving computational resources. However, empirically, a maximum extending length value larger than 6000 would not be helpful for saving computational resources. This value would not be precise in controlling output size, especially when pre-group (followed by '-P') is turn on.In the auto mode, the maximum extending length is estimated based on the sizes of the gap regions that not covered in the seed sequences. A sequence of a closely related species would be preferred for estimating a better maximum extending length value. If you are using limited loci, e.g. rbcL gene as the seed for assembling the whole plastome (with extending length ca. 75000 >> 6000), you should set maximum extending length to inf. Should be a list of numbers/auto/inf split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). Default: inf. ASSEMBLY OPTIONS: These options are about the assembly and graph disentangling -k SPADES_KMER SPAdes kmer settings. Use the same format as in SPAdes. illegal kmer values would be automatically discarded by GetOrganelle. Default: 21,55,85,115 --spades-options OTHER_SPADES_OPTIONS Other SPAdes options. Use double quotation marks to include all the arguments and parameters. --no-spades Disable SPAdes. --ignore-k IGNORE_KMER_RES A kmer threshold below which, no slimming/disentangling would be executed on the result. Default: 40 --genes GENES_FASTA Followed with a customized database (a fasta file or the base name of a blast database) containing or made of ONE set of protein coding genes and ribosomal RNAs extracted from ONE reference genome that you want to assemble. Should be a list of databases split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). This is optional for any organelle mentioned in '-F' but required for 'anonym'. By default, certain database(s) in /nfs1/CGRB/databases/GetOrganelle/latest/LabelDatabase would be used contingent on the organelle types chosen (-F). The default value become invalid when '--genes' or '--ex-genes' is used. --ex-genes EXCLUDE_GENES This is optional and Not suggested, since non-target contigs could contribute information for better downstream coverage-based clustering. Followed with a customized database (a fasta file or the base name of a blast database) containing or made of protein coding genes and ribosomal RNAs extracted from reference genome(s) that you want to exclude. Could be a list of databases split by comma(s) but NOT required to have the same list length to organelle_type (followed by '-F'). The default value will become invalid when '--genes' or '--ex-genes' is used. --disentangle-df DISENTANGLE_DEPTH_FACTOR Depth factor for differentiate genome type of contigs. The genome type of contigs are determined by blast. Default: 10.0 --contamination-depth CONTAMINATION_DEPTH Depth factor for confirming contamination in parallel contigs. Default: 3.0 --contamination-similarity CONTAMINATION_SIMILARITY Similarity threshold for confirming contaminating contigs. Default: 0.9 --no-degenerate Disable making consensus from parallel contig based on nucleotide degenerate table. --degenerate-depth DEGENERATE_DEPTH Depth factor for confirming parallel contigs. Default: 1.5 --degenerate-similarity DEGENERATE_SIMILARITY Similarity threshold for confirming parallel contigs. Default: 0.98 --disentangle-time-limit DISENTANGLE_TIME_LIMIT Time limit (second) for each try of disentangling a graph file as a circular genome. Disentangling a graph as contigs is not limited. Default: 1800 --expected-max-size EXPECTED_MAX_SIZE Expected maximum target genome size(s) for disentangling. Should be a list of INTEGER numbers split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). Default: 250000 (-F embplant_pt/fungus_mt), 25000 (-F embplant_nr/animal_mt/fungus_nr), 1000000 (-F embplant_mt/other_pt),1000000,1000000,250000 (-F other_pt,embplant_mt,fungus_mt) --expected-min-size EXPECTED_MIN_SIZE Expected minimum target genome size(s) for disentangling. Should be a list of INTEGER numbers split by comma(s) on a multi-organelle mode, with the same list length to organelle_type (followed by '-F'). Default: 10000 for all. --reverse-lsc For '-F embplant_pt' with complete circular result, by default, the direction of the starting contig (usually the LSC region) is determined as the direction with less ORFs. Choose this option to reverse the direction of the starting contig when result is circular. Actually, both directions are biologically equivalent to each other. The reordering of the direction is only for easier downstream analysis. --max-paths-num MAX_PATHS_NUM Repeats would dramatically increase the number of potential isomers (paths). This option was used to export a certain amount of paths out of all possible paths per assembly graph. Default: 1000 ADDITIONAL OPTIONS: -t THREADS Maximum threads to use. -P PRE_GROUPED The maximum number (integer) of high-covered reads to be pre-grouped before extending process. pre_grouping is suggested when the whole genome coverage is shallow but the organ genome coverage is deep. The default value is 2E5. For personal computer with 8G memory, we suggest no more than 3E5. A larger number (ex. 6E5) would run faster but exhaust memory in the first few minutes. Choose 0 to disable this process. --which-blast WHICH_BLAST Assign the path to BLAST binary files if not added to the path. Default: try "/local/cluster/getorganelle-1.7.5.0/lib/python3.9/site- packages/GetOrganelleDep/linux/ncbi-blast" first, then $PATH --which-bowtie2 WHICH_BOWTIE2 Assign the path to Bowtie2 binary files if not added to the path. Default: try "/local/cluster/getorganelle-1.7.5.0/lib/python3.9/site- packages/GetOrganelleDep/linux/bowtie2" first, then $PATH --which-spades WHICH_SPADES Assign the path to SPAdes binary files if not added to the path. Default: try "/local/cluster/getorganelle-1.7.5.0/lib/python3.9/site- packages/GetOrganelleDep/linux/SPAdes" first, then $PATH --which-bandage WHICH_BANDAGE Assign the path to bandage binary file if not added to the path. Default: try $PATH --continue Several check points based on files produced, rather than on the log file, so keep in mind that this script will NOT detect the difference between this input parameters and the previous ones. --overwrite Overwrite previous file if existed. --index-in-memory Keep index in memory. Choose save index in memory than in disk. --remove-duplicates RM_DUPLICATES By default this script use unique reads to extend. Choose the number of duplicates (integer) to be saved in memory. A larger number (ex. 2E7) would run faster but exhaust memory in the first few minutes. Choose 0 to disable this process. Note that whether choose or not will not disable the calling of replicate reads. Default: 10000000.0. --flush-step ECHO_STEP Flush step (INTEGER OR INF) for presenting progress. For running in the background, you could set this to inf, which would disable this. Default: 54321 --random-seed RANDOM_SEED Default: 12345 --verbose Verbose output. Choose to enable verbose running log_handler. ``` Please denote the version of GetOrganelle as well as the dependencies in your manuscript for reproducible science. Citation: Jian-Jun Jin*, Wen-Bin Yu*, Jun-Bo Yang, Yu Song, Claude W. dePamphilis, Ting-Shuang Yi, De-Zhu Li. GetOrganelle: a fast and versatile toolkit for accurate de novo assembly of organelle genomes. Genome Biology 21, 241 (2020). [https://doi.org/10.1186/s13059-020-02154-5](https://doi.org/10.1186/s13059-020-02154-5) License: GPL https://www.gnu.org/licenses/gpl-3.0.html Please also cite the dependencies if used: SPAdes: [Bankevich, A., S. Nurk, D. Antipov, A. A. Gurevich, M. Dvorkin, A. S. Kulikov, V. M. Lesin, S. I. Nikolenko, S. Pham, A. D. Prjibelski, A. V. Pyshkin, A. V. Sirotkin, N. Vyahhi, G. Tesler, M. A. Alekseyev and P. A. Pevzner. 2012. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. Journal of Computational Biology 19: 455-477.](https://www.liebertpub.com/doi/abs/10.1089/cmb.2012.0021) Bowtie2: [Langmead, B. and S. L. Salzberg. 2012. Fast gapped-read alignment with Bowtie 2. Nature Methods 9: 357-359.](https://www.nature.com/articles/nmeth.1923) BLAST+: [Camacho, C., G. Coulouris, V. Avagyan, N. Ma, J. Papadopoulos, K. Bealer and T. L. Madden. 2009. BLAST+: architecture and applications. BMC Bioinformatics 10: 421.](https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-10-421) Bandage: [Wick, R. R., M. B. Schultz, J. Zobel and K. E. Holt. 2015. Bandage: interactive visualization of de novo genome assemblies. Bioinformatics 31: 3350-3352.](https://academic.oup.com/bioinformatics/article/31/20/3350/196114) software ref: software ref: research ref: