[Metabuli | Fastp] Init Metabuli standalone workflow

.dockstore.yml

@@ -240,6 +240,11 @@ workflows:
    primaryDescriptorPath: /workflows/phylogenetics/wf_phylocompare.wdl
    testParameterFiles:
     - /tests/inputs/empty.json
+ - name: Metabuli_PHB
+   subclass: WDL
+   primaryDescriptorPath: /workflows/standalone_modules/wf_metabuli.wdl
+   testParameterFiles:
+    - /tests/inputs/empty.json
  - name: Nextclade_Batch_PHB 
    subclass: WDL
    primaryDescriptorPath: /workflows/phylogenetics/wf_nextclade_batch.wdl

docs/assets/tables/all_outputs.tsv

@@ -172,8 +172,8 @@ biosample_metadata	File	Metadata used for BioSample submission in proper BioSamp
 biosample_report_xmls	Array[File]	One or more XML files that contain the response from NCBI regarding your BioSample submission. These can be pretty cryptic, but often contain information to determine  if anything went wrong	Terra_2_NCBI
 biosample_status	String	String showing whether BioSample submission was successful	Terra_2_NCBI
 biosample_submission_xml	File	XML file used to submit your BioSamples to NCBI	Terra_2_NCBI
-bracken_report	String	Refined kraken2 report generated by Bracken	Freyja_FASTQ, TheiaCoV_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE
-bracken_report	File	Refined kraken2 report generated by Bracken	Kraken_PE, Kraken_SE, TheiaCoV_Illumina_SE
+bracken_report	String	Refined kraken2 report generated by Bracken	Freyja_FASTQ, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE
+bracken_report	File	Refined kraken2 report generated by Bracken	Kraken_PE, Kraken_SE
 bracken_report_clean	String	Cleaned refined kraken2 report generated by Bracken	TheiaMeta_Illumina_PE
 bracken_report_dehosted	String	Dehosted refined kraken2 report generated by Bracken	Freyja_FASTQ, TheiaCoV_Illumina_PE
 bracken_report_dehosted	File	Dehosted refined kraken2 report generated by Bracken	TheiaCoV_Illumina_SE
@@ -322,21 +322,28 @@ est_coverage_clean	Float	Estimated coverage calculated from clean reads and geno
 est_coverage_raw	Float	Estimated coverage calculated from raw reads and genome length	TheiaCoV_ONT, TheiaEuk_Illumina_PE, TheiaEuk_ONT, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaProk_ONT
 est_genome_length	Int	Estimated genome length	TheiaEuk_ONT
 est_percent_gene_coverage_tsv	File	Percent coverage for each gene in the organism being analyzed (depending on the organism input)	Freyja_FASTQ, TheiaCoV_ClearLabs, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaCoV_ONT
-ete4_docker	String	Docker image used for ETE4 taxonomy parsing	TheiaViral_Illumina_PE, TheiaViral_ONT
+ete4_docker	String	Docker image used for ETE4 taxonomy parsing	Metabuli, TheiaViral_Illumina_PE, TheiaViral_ONT
 ete4_status	String	Status of standardized taxonomy acquisition	TheiaViral_Illumina_PE, TheiaViral_ONT
-ete4_version	String	The version of ETE4 used	TheiaViral_Illumina_PE, TheiaViral_ONT
+ete4_version	String	The version of ETE4 used	Metabuli, TheiaViral_Illumina_PE, TheiaViral_ONT
 excluded_samples	File	A file that contains the names and reasons why a sample was excluded from submission. **For SARS-CoV-2**, there are two sections: First, a section for any samples that failed to meet pre-determined quality thresholds (`number_N` and `vadr_num_alert`). Second, a section that includes a table that describes any missing required metadata for each sample. This table has the sample name for rows and any columns that have missing metadata as headers. If a sample is missing a piece of required metadata, the corresponding cell will be blank. However, if a different sample does have metadata for that column, the associated value will appear in the corresponding cell. **For flu and mpox**, only the second section described above exists. _Please see the example below for more details_.	Mercury_Prep_N_Batch
 excluded_samples	File	Text file listing samples that were excluded from BioSample submission for missing required metadata	Terra_2_NCBI
 failed_uploads	File	The metadata for any failed uploads	Terra_2_GISAID
 fasta_utilities_fai	File	FASTA index file	TheiaViral_ONT
 fasta_utilities_samtools_docker	String	Docker image used for samtools in fasta utilities	TheiaViral_ONT
 fasta_utilities_samtools_version	String	Version of samtools used in fasta utilities	TheiaViral_ONT
-fastp_docker	String	Docker image used for fastp	Freyja_FASTQ, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
-fastp_html_report	File	The HTML report made with fastp	TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
-fastp_html_report	String	The HTML report made with fastp	Freyja_FASTQ
-fastp_json_report	File	The JSON report made with fastp	TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
-fastp_json_report	String	The JSON report made with fastp	Freyja_FASTQ
-fastp_version	String	The version of fastp used	Freyja_FASTQ, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
+fastp_docker	String	Docker image used for fastp	Freyja_FASTQ, Metabuli, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
+fastp_html_report	File	The HTML report conveying fastp results	Metabuli, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
+fastp_html_report	String	The HTML report conveying fastp results	Freyja_FASTQ
+fastp_json_report	File	The JSON report conveying fastp results	Metabuli, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
+fastp_json_report	String	The JSON report conveying fastp results	Freyja_FASTQ
+fastp_read1_trimmed	File	`read1` input trimmed by fastp	Metabuli
+fastp_read2_trimmed	File	`read2` input trimmed by fastp	Metabuli
+fastp_version	String	The version of fastp used	Freyja_FASTQ, Metabuli, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaEuk_Illumina_PE, TheiaMeta_Illumina_PE, TheiaProk_Illumina_PE, TheiaProk_Illumina_SE, TheiaViral_Illumina_PE, TheiaViral_Panel
+fastplong_docker	String	Docker image used for fastplong	Metabuli
+fastplong_html_report	File	The HTML report conveying fastplong results	Metabuli
+fastplong_json_report	File	The JSON report conveying fastplong results	Metabuli
+fastplong_read1_trimmed	File	`read1` input trimmed by fastplong	Metabuli
+fastplong_version	String	The version of fastplong used	Metabuli
 fastq_dl_date	String	The date of the read data download	SRA_Fetch
 fastq_dl_docker	String	The docker used	SRA_Fetch
 fastq_dl_fastq_metadata	File	File containing metadata of the provided accession such as submission_accession, library_selection, instrument_platform, among others	SRA_Fetch
@@ -673,21 +680,26 @@ mercury_prep_n_batch_version	String	Version of the PHB repository that hosts thi
 mercury_script_version	String	Version of the Mercury tool that was used in this workflow	Mercury_Prep_N_Batch
 merlin_tag	String	Merlin tag for the assembly	TheiaEuk_ONT
 metabuli_classified	File	Classified reads from Metabuli	TheiaViral_ONT
+metabuli_classified_read1	File	FASTQ of `read1` input classified by Metabuli	Metabuli
+metabuli_classified_read2	File	FASTQ of `read2` input classified by Metabuli	Metabuli
+metabuli_classified_report	File	Classification report from Metabuli	Metabuli
 metabuli_database	String	Database used for Metabuli	TheiaProk_ONT, TheiaViral_ONT
-metabuli_docker	String	Docker image used for Metabuli	TheiaProk_ONT, TheiaViral_ONT
+metabuli_docker	String	Docker image used for Metabuli	Metabuli, TheiaProk_ONT, TheiaViral_ONT
 metabuli_human	Float	Percent of human reads detected in raw reads	Freyja_FASTQ, TheiaCoV_ONT
 metabuli_human_dehosted	Float	Percent of human reads detected after removing human reads	Freyja_FASTQ, TheiaCoV_ONT
-metabuli_krona_report	File	Krona visualization report from Metabuli	TheiaViral_ONT
+metabuli_krona_report	File	Krona visualization report from Metabuli	Metabuli, TheiaViral_ONT
 metabuli_percent_target_lineage	String	Percent of reads classified to the target lineage	TheiaViral_ONT
 metabuli_read1_extract	File	Extracted reads from Metabuli	TheiaViral_ONT
 metabuli_report	String	Classification report from Metabuli	Freyja_FASTQ, TheiaCoV_ONT, TheiaProk_ONT
-metabuli_report	File	Classification report from Metabuli	TheiaViral_ONT
+metabuli_report	File	Classification report from Metabuli	Metabuli, TheiaViral_ONT
 metabuli_report_dehosted	String	Classification report from Metabuli after removing human reads	Freyja_FASTQ, TheiaCoV_ONT
-metabuli_status	String	Status of Metabuli analysis	TheiaViral_ONT
+metabuli_status	String	Status of Metabuli analysis	Metabuli, TheiaViral_ONT
 metabuli_target_organism	String	Percent of reads classified to target organism for Metabuli analysis	TheiaCoV_ONT
 metabuli_target_organism_dehosted	String	Percent of reads classified to target organism for Metabuli analysis after removing human reads	TheiaCoV_ONT
 metabuli_target_organism_name	String	Name of the target organism for Metabuli analysis	TheiaCoV_ONT
-metabuli_version	String	Version of Metabuli used	Freyja_FASTQ, TheiaCoV_ONT, TheiaProk_ONT, TheiaViral_ONT
+metabuli_version	String	Version of Metabuli used	Freyja_FASTQ, Metabuli, TheiaCoV_ONT, TheiaProk_ONT, TheiaViral_ONT
+metabuli_wf_analysis_date	String	Analysis date for Metabuli workflow	Metabuli
+metabuli_wf_version	String	Version of Metabuli workflow	Metabuli
 metaspades_docker	String	The Docker image of metaspades	TheiaMeta_Illumina_PE
 metaspades_version	String	The version of metaspades	TheiaMeta_Illumina_PE
 metaviralspades_docker	String	Docker image used for MetaviralSPAdes	TheiaViral_Illumina_PE
@@ -730,7 +742,7 @@ nanoplot_tsv_raw	File	A TSV report describing the raw reads	Freyja_FASTQ, TheiaC
 nanoplot_version	String	Version of nanoplot used for analysis	TheiaCoV_ONT, TheiaEuk_ONT, TheiaProk_ONT
 nanoq_filtered_read1	File	Filtered reads from NanoQ	TheiaViral_ONT
 nanoq_version	String	Version of nanoq used in analysis	Freyja_FASTQ, TheiaEuk_ONT, TheiaProk_ONT, TheiaViral_ONT
-ncbi_read_extraction_rank	String	Read extraction rank used	TheiaViral_Illumina_PE, TheiaViral_ONT
+ncbi_read_extraction_rank	String	Read extraction rank used	Metabuli, TheiaViral_Illumina_PE, TheiaViral_ONT
 ncbi_scrub_docker	String	The Docker image for NCBI's HRRT (human read removal tool)	NCBI_Scrub_PE, NCBI_Scrub_SE, TheiaMeta_Illumina_PE, TheiaViral_Illumina_PE, TheiaViral_ONT, TheiaViral_Panel
 ncbi_scrub_human_spots_removed	Int	Number of spots removed (or masked)	NCBI_Scrub_PE, NCBI_Scrub_SE, TheiaViral_Illumina_PE, TheiaViral_ONT, TheiaViral_Panel
 ncbi_scrub_pe_analysis_date	String	Date of analysis	NCBI_Scrub_PE
@@ -739,8 +751,8 @@ ncbi_scrub_read1_dehosted	File	Dehosted reads after NCBI scrub	TheiaViral_Illumi
 ncbi_scrub_read2_dehosted	File	Dehosted reads after NCBI scrub	TheiaViral_Illumina_PE, TheiaViral_Panel
 ncbi_scrub_se_analysis_date	String	Date of analysis	NCBI_Scrub_SE
 ncbi_scrub_se_version	String	Version of HRRT software used	NCBI_Scrub_SE
-ncbi_taxon_id	String	NCBI taxonomy ID of inputted organism following rank extraction	TheiaViral_Illumina_PE, TheiaViral_ONT
-ncbi_taxon_name	String	NCBI taxonomy name of inputted taxon following rank extraction	TheiaViral_Illumina_PE, TheiaViral_ONT
+ncbi_taxon_id	String	NCBI taxonomy ID of inputted organism following rank extraction	Metabuli, TheiaViral_Illumina_PE, TheiaViral_ONT
+ncbi_taxon_name	String	NCBI taxonomy name of inputted taxon following rank extraction	Metabuli, TheiaViral_Illumina_PE, TheiaViral_ONT
 nextclade_aa_dels	String	Amino-acid deletions as detected by NextClade. Will be blank for Flu	TheiaCoV_ClearLabs, TheiaCoV_FASTA, TheiaCoV_Illumina_PE, TheiaCoV_Illumina_SE, TheiaCoV_ONT, TheiaViral_Illumina_PE, TheiaViral_ONT
 nextclade_aa_dels_flu_h5n1	String	Amino-acid deletions as detected by NextClade. Specific to flu; it includes deletions for H5N1 whole genome	TheiaCoV_FASTA, TheiaCoV_Illumina_PE, TheiaCoV_ONT
 nextclade_aa_dels_flu_ha	String	Amino-acid deletions as detected by NextClade. Specific to flu; it includes deletions for HA segment	TheiaCoV_FASTA, TheiaCoV_Illumina_PE, TheiaCoV_ONT, TheiaViral_Illumina_PE, TheiaViral_ONT