kkonganti@2: <tool id="hfp_nowayout" name="nowayout" version="0.5.0+galaxy24">
kkonganti@0:     <description>An automated workflow to identify Mitochondrial reads and classify Eukaryotes.</description>
kkonganti@0:     <requirements>
kkonganti@0:         <container type="docker">quay.io/biocontainers/nextflow:24.10.4--hdfd78af_0</container>
kkonganti@0:     </requirements> 
kkonganti@0:     <version_command>nextflow -version</version_command>
kkonganti@0:     <command detect_errors="exit_code"><![CDATA[
kkonganti@0: 	input_path=\$(pwd)"/cpipes-input";
kkonganti@0:     mkdir -p "\${input_path}" || exit 1;
kkonganti@0:     #import re
kkonganti@0:     #if (str($input_read_type_cond.input_read_type) == "single_long"):
kkonganti@0: 	    #for _, $unpaired in enumerate($input_read_type_cond.input):
kkonganti@0:             #set read1 = str($unpaired.name)
kkonganti@0:             #if not str($unpaired.name).endswith(('.fastq', '.fastq.gz')):
kkonganti@0:                 #set read1_ext = re.sub('fastqsanger', 'fastq', str($unpaired.ext))
kkonganti@0:                 #set read1 = str($unpaired.name) + str('.') + $read1_ext
kkonganti@0:             #end if
kkonganti@0:             ln -sf '$unpaired' "\${input_path}/$read1";
kkonganti@0: 	    #end for
kkonganti@0:     #elif (str($input_read_type_cond.input_read_type) == "paired"):
kkonganti@0: 	    #for _, $pair in enumerate($input_read_type_cond.input_pair)
kkonganti@0:             #set read_R1 = re.sub('\:forward', '_forward', str($pair.forward.name))
kkonganti@0:             #set read_R2 = re.sub('\:reverse', '_reverse', str($pair.reverse.name))
kkonganti@0:             #set read_R1_ext = re.sub('fastqsanger', 'fastq', str($pair.forward.ext))
kkonganti@0:             #set read_R2_ext = re.sub('fastqsanger', 'fastq', str($pair.reverse.ext))
kkonganti@0:             #if not str($pair.forward.name).endswith(('.fastq', '.fastq.gz')):
kkonganti@0: 	            #set read_R1 = $read_R1 + str('.') + $read_R1_ext
kkonganti@0:             #end if
kkonganti@0:             #if not str($pair.reverse.name).endswith(('.fastq', '.fastq.gz')):
kkonganti@0:                 #set read_R2 = $read_R2 + str('.') + $read_R2_ext
kkonganti@0:             #end if
kkonganti@0: 	        ln -sf '$pair.forward' "\${input_path}/$read_R1";
kkonganti@0: 	        ln -sf '$pair.reverse' "\${input_path}/$read_R2";
kkonganti@0: 	    #end for
kkonganti@0:     #end if
kkonganti@0: 	$__tool_directory__/0.5.0/cpipes
kkonganti@0:     --pipeline nowayout
kkonganti@0:     --input \${input_path}
kkonganti@0: 	--output cpipes-output
kkonganti@0:     --fq_suffix '${input_read_type_cond.fq_suffix}'
kkonganti@0:     #if (str($input_read_type_cond.input_read_type) == "single_long"):
kkonganti@0:         --fq_single_end true
kkonganti@0:     #elif (str($input_read_type_cond.input_read_type) == "paired"):
kkonganti@0:         --fq_single_end false --fq2_suffix '${input_read_type_cond.fq2_suffix}'
kkonganti@0:     #end if
kkonganti@0:     --db_mode $nowo_db_mode
kkonganti@0:     --nowo_thresholds $nowo_thresholds
kkonganti@0: 	--fq_filename_delim '${fq_filename_delim}'
kkonganti@0: 	--fq_filename_delim_idx $fq_filename_delim_idx
kkonganti@0: 	-profile gxkubernetes;
kkonganti@0:     mv './cpipes-output/nowayout-multiqc/multiqc_report.html' './multiqc_report.html' || exit 1;
kkonganti@0:     mv './cpipes-output/krona_ktimporttext/CPIPES_nowayout_krona.html './CPIPES_nowayout_krona.html' || exit 1;
kkonganti@0:     rm -rf ./cpipes-output || exit 1;
kkonganti@0:     rm -rf ./work || exit 1;
kkonganti@0:     ]]></command>
kkonganti@0:     <inputs>
kkonganti@0:         <conditional name="input_read_type_cond">
kkonganti@0:             <param name="input_read_type" type="select" label="Select the read collection type">
kkonganti@0:                 <option value="single_long" selected="true">Single-End short reads</option>
kkonganti@0:                 <option value="paired">Paired-End short reads</option>
kkonganti@0:             </param>
kkonganti@0:             <when value="single_long">
kkonganti@0:                 <param name="input" type="data_collection" collection_type="list" format="fastq,fastq.gz"
kkonganti@0:                     label="Dataset list of unpaired short reads or long reads" />
kkonganti@0:                 <param name="fq_suffix" value=".fastq.gz" type="text" label="Suffix of the Single-End FASTQ"/>
kkonganti@0:             </when>
kkonganti@0:             <when value="paired">
kkonganti@0:                 <param name="input_pair" type="data_collection" collection_type="list:paired" format="fastq,fastq.gz" label="List of Dataset pairs" />
kkonganti@0:                 <param name="fq_suffix" value="_R1_001.fastq.gz" type="text" label="Suffix of the R1 FASTQ"
kkonganti@0:                     help="For any data sets downloaded from NCBI into Galaxy, change this to _forward.fastq.gz suffix."/>
kkonganti@0:                 <param name="fq2_suffix" value="_R2_001.fastq.gz" type="text" label="Suffix of the R2 FASTQ"
kkonganti@0:                     help="For any data sets downloaded from NCBI into Galaxy, change this to _reverse.fastq.gz suffix."/>
kkonganti@0:             </when>
kkonganti@0:         </conditional>
kkonganti@0:         <param name="nowo_db_mode" type="select" label="Select the database with nowayout"
kkonganti@0:             help="Please see below about different databases.">
kkonganti@0:             <option value="mitomine" selected="true">mitomine</option>
kkonganti@0:             <option value="cytox1">cytox1</option>
kkonganti@0:             <option value="voucher">voucher</option>
kkonganti@0:             <option value="ganoderma">ganoderma</option>
kkonganti@0:             <option value="listeria">listeria</option>
kkonganti@0:         </param>
kkonganti@0:         <param name="nowo_thresholds" type="select" label="Enter the type of base quality thresholds to be set with nowayout"
kkonganti@0:             help="The default value sets strictest thresholds that tends to filter out most of the false positive hits.">
kkonganti@0:             <option value="strict" selected="true">strict</option>
kkonganti@0:             <option value="relax">relax</option>
kkonganti@0:         </param>
kkonganti@0:         <param name="fq_filename_delim" type="text" value="_" label="File name delimitor by which samples are grouped together (--fq_filename_delim)" 
kkonganti@0:             help="This is the delimitor by which samples are grouped together to display in the final MultiQC report. For example, if your input data sets are mango_replicate1.fastq.gz, mango_replicate2.fastq.gz, orange_replicate1_maryland.fastq.gz, orange_replicate2_maryland.fastq.gz, then to create 2 samples mango and orange, the value for --fq_filename_delim would be _ (underscore) and the value for --fq_filename_delim_idx would be 1, since you want to group by the first word (i.e. mango or orange) after splitting the filename based on _ (underscore)."/>
kkonganti@0:         <param name="fq_filename_delim_idx" type="integer" value="1" label="File name delimitor index (--fq_filename_delim_idx)" />
kkonganti@0:     </inputs>
kkonganti@0:     <outputs>
kkonganti@0:         <data name="krona_chart" format="html" label="nowayout: Krona Chart on ${on_string}" from_work_dir="CPIPES_nowayout_krona.html"/>
kkonganti@0:         <data name="multiqc_report" format="html" label="nowayout: MultiQC Report on ${on_string}" from_work_dir="multiqc_report.html"/>
kkonganti@0:     </outputs>
kkonganti@0:     <tests>
kkonganti@0:         <!--Test 01: long reads-->
kkonganti@0:         <test expect_num_outputs="2">
kkonganti@0:             <param name="input">
kkonganti@0:                 <collection type="list">
kkonganti@0:                     <element name="FAL11127.fastq.gz" value="FAL11127.fastq.gz" />
kkonganti@0:                     <element name="FAL11341.fastq.gz" value="FAL11341.fastq.gz" />
kkonganti@0:                     <element name="FAL11342.fastq.gz" value="FAL11342.fastq.gz" />
kkonganti@0:                 </collection>
kkonganti@0:             </param>
kkonganti@0:             <param name="fq_suffix" value=".fastq.gz"/>
kkonganti@0:             <output name="multiqc_report" file="multiqc_report.html" ftype="html" compare="sim_size"/>
kkonganti@0:             <!-- <output name="assembled_mags" file="FAL11127.assembly_filtered.contigs.fasta" ftype="fasta" compare="sim_size"/> -->
kkonganti@0:         </test>
kkonganti@0:     </tests>
kkonganti@0:     <help><![CDATA[
kkonganti@0: 
kkonganti@0: .. class:: infomark
kkonganti@0: 
kkonganti@0: **Purpose**
kkonganti@0: 
kkonganti@0: nowayout is a mitochondrial metagenomics classifier for Eukaryotes. 
kkonganti@0: It uses a custom kma database to identify mitochondrial reads and 
kkonganti@0: performs read classification followed by further read classification 
kkonganti@0: reinforcement using sourmash.
kkonganti@0: 
kkonganti@0: It is written in Nextflow and is part of the modular data analysis pipelines (CFSAN PIPELINES or CPIPES for short) at HFP.
kkonganti@0: 
kkonganti@0: 
kkonganti@0: ----
kkonganti@0: 
kkonganti@0: .. class:: infomark
kkonganti@0: 
kkonganti@3: **Databases**
kkonganti@0: 
kkonganti@3:     - *mitomine*: Big database that works in almost all scenarios.
kkonganti@3:     - *cytox1*: Collection of only non-redundant COXI genes from NCBI.
kkonganti@3:     - *voucher*: Collection of only non-redundant voucher sequences from NCBI.
kkonganti@3:     - *ganoderma*: Collection of only non-redundant mtDNA sequences of Ganoderma fungi.
kkonganti@3:     - *listeria*: Collection of organelle sequences and other rRNA genes for Listeria.
kkonganti@0: 
kkonganti@0: 
kkonganti@0: ----
kkonganti@0: 
kkonganti@0: .. class:: infomark
kkonganti@0: 
kkonganti@0: **Testing and Validation**
kkonganti@0: 
kkonganti@0: The CPIPES - nowayout Nextflow pipeline has been wrapped to make it work in Galaxy. 
kkonganti@0: It takes in either paired or unpaired short reads list as an input and generates a MultiQC report 
kkonganti@0: which contains relative abundances in context of number of mitochondrial reads identified. It also 
kkonganti@0: generates a Krona chart for each sample. The pipeline has been tested on multiple internal insect 
kkonganti@0: mixture samples. All the original testing and validation was done on the command line on the 
kkonganti@0: HFP Reedling HPC Cluster.
kkonganti@0: 
kkonganti@0: 
kkonganti@0: ----
kkonganti@0: 
kkonganti@0: .. class:: infomark
kkonganti@0: 
kkonganti@3: **Please note**
kkonganti@0: 
kkonganti@3:     - *nowayout* only works on Illumina short reads (paired or unpaired).
kkonganti@3:     - *nowayout* uses a custom kma database named *mitomine*.
kkonganti@3:     - The custom database will be incrementally augmented and refined over time.
kkonganti@3:     - *mitomine* stats:
kkonganti@3:         Contains ~ 2.93M non-redundant mitochondrial and voucher sequences.
kkonganti@3:         Represents ~ 717K unique species.
kkonganti@3:     - Other databases are also available but will be seldom updated.
kkonganti@0: 
kkonganti@0: ----
kkonganti@0: 
kkonganti@0: .. class:: infomark
kkonganti@0: 
kkonganti@0: **Outputs**
kkonganti@0: 
kkonganti@0: The main output file is a:
kkonganti@0: 
kkonganti@0:     ::
kkonganti@0: 
kkonganti@0:         - MultiQC Report: Contains a brief summary report including individual Mitochondrial reads identified
kkonganti@0:                           per sample and relative abundances in context of the total number of Mitochondrial reads
kkonganti@0:                           identified.
kkonganti@0:                           Please note that due to MultiQC customizations, the preview (eye icon) will not
kkonganti@0:                           work within Galaxy for the MultiQC report. Please download the file by clicking
kkonganti@0:                           on the floppy icon and view it in your browser on your local desktop/workstation.
kkonganti@0:                           You can export the tables and plots from the downloaded MultiQC report.
kkonganti@0: 
kkonganti@0:   ]]></help>
kkonganti@0:     <citations>
kkonganti@0:         <citation type="bibtex">
kkonganti@0:             @article{nowayout,
kkonganti@0:             author = {Konganti, Kranti},
kkonganti@0:             year = {2025},
kkonganti@0:             month = {May},
kkonganti@0:             title = {nowayout: An automated mitrochiondrial read classifier for Eukaryotes.},
kkonganti@0:             journal = {Manuscript in preparation},
kkonganti@0:             doi = {10.3389/xxxxxxxxxxxxxxxxxx},
kkonganti@0:             url = {https://xxxxxxx/articles/10.3389/xxxxxxxxxxxx/full}}
kkonganti@0:         </citation>
kkonganti@0:     </citations>
kkonganti@0: </tool>