[automated] Create package recipe from SSF files

packages/2023_Salazar_MachuPicchu/2023_Salazar_MachuPicchu.config

@@ -0,0 +1,40 @@
+// Keep track of config versions
+config_template_version='0.3.0dev'
+package_config_version='0.3.0dev'
+minotaur_config_base='https://raw.githubusercontent.com/poseidon-framework/poseidon-eager/main/conf'
+
+// This configuration file is designed to be a used with the nf-core/eager pipeline.
+//   Instead of having to specify all other configurations for the Minotaur pipeline
+//   on runtime, they are all contained in this file and loaded automatically upon
+//   specifying this config file during runtime. Additionally, any parameters that
+//   need to be altered from the defaults can be specified here.
+//
+// The intention is to make it easy for users to understand and reproduce the output
+//  from processing with the Minotaur workflow processing from the contents of a
+//  single file.
+
+// Load configuration profiles. They are loaded from the minotaur_config_base URL, main branch.
+includeConfig "${minotaur_config_base}/EVA_cluster.config" // Cluster-specific configurations for nf-core/eager execution at MPI-EVA
+includeConfig "${minotaur_config_base}/Minotaur.config"    // Default nf-core/eager parameters for Minotaur processing.
+
+// The following config file specifies BED files for on-target endogenous DNA calculation and mean coverage as well as pseudohaploid genotyping.
+// TODO: Select the appropriate config for the CaptureType of the package.
+includeConfig "${minotaur_config_base}/CaptureType_profiles/1240K.config"
+
+params {
+  // Keep track of config file versions used when processing
+  config_profile_description = "${config_profile_description}\n - config_template_version: ${config_template_version}\n - package_config_version: ${package_config_version}"
+  config_profile_contact     = "Thiseas C. Lamnidis (@TCLamnidis)"
+
+  /* 
+  TODO: If you need to change any of the default processing parameters for this package
+  you can specify these parameters below.
+  Any parameters not specified in any of the config files default to their nf-core/eager default values.
+
+  For information on all available parameters and their default values see: 
+    https://nf-co.re/eager/2.4.6/parameters
+
+  You can see the default values for parameters within poseidon-eager at: 
+    https://github.com/poseidon-framework/poseidon-eager/blob/main/conf/Minotaur.config
+  */
+}

packages/2023_Salazar_MachuPicchu/2023_Salazar_MachuPicchu.tsv

@@ -0,0 +1,57 @@
+Sample_Name	Library_ID	Lane	Colour_Chemistry	SeqType	Organism	Strandedness	UDG_Treatment	R1	R2	BAM	R1_target_file	R2_target_file
+KCA-3089	KCA-3089_KCA-3089	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KCA-3089_KCA-3089_L1_R1.fastq.gz	NA	NA	ERR11518429.fastq.gz	NA
+KCA-1108	KCA-1108_KCA-1108	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KCA-1108_KCA-1108_L1_R1.fastq.gz	NA	NA	ERR11518434.fastq.gz	NA
+KCA-1364	KCA-1364_KCA-1364	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KCA-1364_KCA-1364_L1_R1.fastq.gz	NA	NA	ERR11518435.fastq.gz	NA
+MP13	MP13_MP13	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP13_MP13_L1_R1.fastq.gz	<PATH_TO_DATA>/MP13_MP13_L1_R2.fastq.gz	NA	ERR11518444.fastq.gz	ERR11518444_1.fastq.gz
+KMA-28-1	KMA-28-1_KMA-28-1	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KMA-28-1_KMA-28-1_L1_R1.fastq.gz	NA	NA	ERR11518463.fastq.gz	NA
+MP33	MP33_MP33	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP33_MP33_L1_R1.fastq.gz	<PATH_TO_DATA>/MP33_MP33_L1_R2.fastq.gz	NA	ERR11518465.fastq.gz	ERR11518465_1.fastq.gz
+MP42B	MP42B_MP42B	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP42B_MP42B_L1_R1.fastq.gz	<PATH_TO_DATA>/MP42B_MP42B_L1_R2.fastq.gz	NA	ERR11518471.fastq.gz	ERR11518471_1.fastq.gz
+MP42A	MP42A_MP42A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP42A_MP42A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP42A_MP42A_L1_R2.fastq.gz	NA	ERR11518472.fastq.gz	ERR11518472_1.fastq.gz
+MP48B	MP48B_MP48B	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP48B_MP48B_L1_R1.fastq.gz	<PATH_TO_DATA>/MP48B_MP48B_L1_R2.fastq.gz	NA	ERR11518477.fastq.gz	ERR11518477_1.fastq.gz
+MP50A	MP50A_MP50A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP50A_MP50A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP50A_MP50A_L1_R2.fastq.gz	NA	ERR11518484.fastq.gz	ERR11518484_1.fastq.gz
+MP4f	MP4f_MP4f	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP4f_MP4f_L1_R1.fastq.gz	<PATH_TO_DATA>/MP4f_MP4f_L1_R2.fastq.gz	NA	ERR11518485.fastq.gz	ERR11518485_1.fastq.gz
+MP4i	MP4i_MP4i	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP4i_MP4i_L1_R1.fastq.gz	<PATH_TO_DATA>/MP4i_MP4i_L1_R2.fastq.gz	NA	ERR11518489.fastq.gz	ERR11518489_1.fastq.gz
+MP5A	MP5A_MP5A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP5A_MP5A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP5A_MP5A_L1_R2.fastq.gz	NA	ERR11518495.fastq.gz	ERR11518495_1.fastq.gz
+MP65B	MP65B_MP65B	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP65B_MP65B_L1_R1.fastq.gz	<PATH_TO_DATA>/MP65B_MP65B_L1_R2.fastq.gz	NA	ERR11518496.fastq.gz	ERR11518496_1.fastq.gz
+MP27	MP27_MP27	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP27_MP27_L1_R1.fastq.gz	<PATH_TO_DATA>/MP27_MP27_L1_R2.fastq.gz	NA	ERR11518499.fastq.gz	ERR11518499_1.fastq.gz
+OTT-1144	OTT-1144_OTT-1144	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/OTT-1144_OTT-1144_L1_R1.fastq.gz	NA	NA	ERR11518523.fastq.gz	NA
+MP61	MP61_MP61	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP61_MP61_L1_R1.fastq.gz	<PATH_TO_DATA>/MP61_MP61_L1_R2.fastq.gz	NA	ERR11518546.fastq.gz	ERR11518546_1.fastq.gz
+QKI-21-1	QKI-21-1_QKI-21-1	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/QKI-21-1_QKI-21-1_L1_R1.fastq.gz	NA	NA	ERR11518553.fastq.gz	NA
+MP31A	MP31A_MP31A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP31A_MP31A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP31A_MP31A_L1_R2.fastq.gz	NA	ERR11518458.fastq.gz	ERR11518458_1.fastq.gz
+KMA-15A-1	KMA-15A-1_KMA-15A-1	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KMA-15A-1_KMA-15A-1_L1_R1.fastq.gz	NA	NA	ERR11518461.fastq.gz	NA
+MP45A	MP45A_MP45A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP45A_MP45A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP45A_MP45A_L1_R2.fastq.gz	NA	ERR11518475.fastq.gz	ERR11518475_1.fastq.gz
+MP4E	MP4E_MP4E	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP4E_MP4E_L1_R1.fastq.gz	<PATH_TO_DATA>/MP4E_MP4E_L1_R2.fastq.gz	NA	ERR11518480.fastq.gz	ERR11518480_1.fastq.gz
+MP4D	MP4D_MP4D	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP4D_MP4D_L1_R1.fastq.gz	<PATH_TO_DATA>/MP4D_MP4D_L1_R2.fastq.gz	NA	ERR11518481.fastq.gz	ERR11518481_1.fastq.gz
+MP53A	MP53A_MP53A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP53A_MP53A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP53A_MP53A_L1_R2.fastq.gz	NA	ERR11518486.fastq.gz	ERR11518486_1.fastq.gz
+MP55	MP55_MP55	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP55_MP55_L1_R1.fastq.gz	<PATH_TO_DATA>/MP55_MP55_L1_R2.fastq.gz	NA	ERR11518487.fastq.gz	ERR11518487_1.fastq.gz
+MP78A	MP78A_MP78A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP78A_MP78A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP78A_MP78A_L1_R2.fastq.gz	NA	ERR11518501.fastq.gz	ERR11518501_1.fastq.gz
+MP84A	MP84A_MP84A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP84A_MP84A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP84A_MP84A_L1_R2.fastq.gz	NA	ERR11518514.fastq.gz	ERR11518514_1.fastq.gz
+QKI-809	QKI-809_QKI-809	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/QKI-809_QKI-809_L1_R1.fastq.gz	NA	NA	ERR11518566.fastq.gz	NA
+QKI-9-3	QKI-9-3_QKI-9-3	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/QKI-9-3_QKI-9-3_L1_R1.fastq.gz	NA	NA	ERR11518572.fastq.gz	NA
+SHN-575	SHN-575_SHN-575	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/SHN-575_SHN-575_L1_R1.fastq.gz	NA	NA	ERR11518597.fastq.gz	NA
+KCA-1298	KCA-1298_KCA-1298	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KCA-1298_KCA-1298_L1_R1.fastq.gz	NA	NA	ERR11518432.fastq.gz	NA
+CCA-7-2	CCA-7-2_CCA-7-2	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/CCA-7-2_CCA-7-2_L1_R1.fastq.gz	NA	NA	ERR11518440.fastq.gz	NA
+MP107B	MP107B_MP107B	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP107B_MP107B_L1_R1.fastq.gz	<PATH_TO_DATA>/MP107B_MP107B_L1_R2.fastq.gz	NA	ERR11518443.fastq.gz	ERR11518443_1.fastq.gz
+MP23	MP23_MP23	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP23_MP23_L1_R1.fastq.gz	<PATH_TO_DATA>/MP23_MP23_L1_R2.fastq.gz	NA	ERR11518448.fastq.gz	ERR11518448_1.fastq.gz
+KMA-2-2	KMA-2-2_KMA-2-2	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KMA-2-2_KMA-2-2_L1_R1.fastq.gz	NA	NA	ERR11518456.fastq.gz	NA
+MP32	MP32_MP32	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP32_MP32_L1_R1.fastq.gz	<PATH_TO_DATA>/MP32_MP32_L1_R2.fastq.gz	NA	ERR11518466.fastq.gz	ERR11518466_1.fastq.gz
+KMA-19-1	KMA-19-1_KMA-19-1	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/KMA-19-1_KMA-19-1_L1_R1.fastq.gz	NA	NA	ERR11518467.fastq.gz	NA
+MP42C	MP42C_MP42C	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP42C_MP42C_L1_R1.fastq.gz	<PATH_TO_DATA>/MP42C_MP42C_L1_R2.fastq.gz	NA	ERR11518469.fastq.gz	ERR11518469_1.fastq.gz
+MP4B	MP4B_MP4B	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP4B_MP4B_L1_R1.fastq.gz	<PATH_TO_DATA>/MP4B_MP4B_L1_R2.fastq.gz	NA	ERR11518478.fastq.gz	ERR11518478_1.fastq.gz
+MP51	MP51_MP51	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP51_MP51_L1_R1.fastq.gz	<PATH_TO_DATA>/MP51_MP51_L1_R2.fastq.gz	NA	ERR11518493.fastq.gz	ERR11518493_1.fastq.gz
+MP71	MP71_MP71	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP71_MP71_L1_R1.fastq.gz	<PATH_TO_DATA>/MP71_MP71_L1_R2.fastq.gz	NA	ERR11518498.fastq.gz	ERR11518498_1.fastq.gz
+MP77A	MP77A_MP77A	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP77A_MP77A_L1_R1.fastq.gz	<PATH_TO_DATA>/MP77A_MP77A_L1_R2.fastq.gz	NA	ERR11518500.fastq.gz	ERR11518500_1.fastq.gz
+MP82	MP82_MP82	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP82_MP82_L1_R1.fastq.gz	<PATH_TO_DATA>/MP82_MP82_L1_R2.fastq.gz	NA	ERR11518502.fastq.gz	ERR11518502_1.fastq.gz
+MP80	MP80_MP80	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP80_MP80_L1_R1.fastq.gz	<PATH_TO_DATA>/MP80_MP80_L1_R2.fastq.gz	NA	ERR11518503.fastq.gz	ERR11518503_1.fastq.gz
+MP84C	MP84C_MP84C	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP84C_MP84C_L1_R1.fastq.gz	<PATH_TO_DATA>/MP84C_MP84C_L1_R2.fastq.gz	NA	ERR11518512.fastq.gz	ERR11518512_1.fastq.gz
+OTT-1160	OTT-1160_OTT-1160	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/OTT-1160_OTT-1160_L1_R1.fastq.gz	NA	NA	ERR11518538.fastq.gz	NA
+OTT-1165	OTT-1165_OTT-1165	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/OTT-1165_OTT-1165_L1_R1.fastq.gz	NA	NA	ERR11518543.fastq.gz	NA
+MP63	MP63_MP63	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP63_MP63_L1_R1.fastq.gz	<PATH_TO_DATA>/MP63_MP63_L1_R2.fastq.gz	NA	ERR11518550.fastq.gz	ERR11518550_1.fastq.gz
+OTT-1174	OTT-1174_OTT-1174	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/OTT-1174_OTT-1174_L1_R1.fastq.gz	NA	NA	ERR11518552.fastq.gz	NA
+QKI-4	QKI-4_QKI-4	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/QKI-4_QKI-4_L1_R1.fastq.gz	NA	NA	ERR11518562.fastq.gz	NA
+OTT-1187	OTT-1187_OTT-1187	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/OTT-1187_OTT-1187_L1_R1.fastq.gz	NA	NA	ERR11518565.fastq.gz	NA
+SHN-212	SHN-212_SHN-212	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/SHN-212_SHN-212_L1_R1.fastq.gz	NA	NA	ERR11518568.fastq.gz	NA
+SHN-355	SHN-355_SHN-355	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/SHN-355_SHN-355_L1_R1.fastq.gz	NA	NA	ERR11518578.fastq.gz	NA
+SHN-492	SHN-492_SHN-492	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/SHN-492_SHN-492_L1_R1.fastq.gz	NA	NA	ERR11518582.fastq.gz	NA
+SHN-902	SHN-902_SHN-902	1	4	SE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/SHN-902_SHN-902_L1_R1.fastq.gz	NA	NA	ERR11518584.fastq.gz	NA
+MP3a	MP3a_MP3a	1	4	PE	Homo sapiens (modern human)	double	half	<PATH_TO_DATA>/MP3a_MP3a_L1_R1.fastq.gz	<PATH_TO_DATA>/MP3a_MP3a_L1_R2.fastq.gz	NA	ERR11518632.fastq.gz	ERR11518632_1.fastq.gz

packages/2023_Salazar_MachuPicchu/2023_Salazar_MachuPicchu.tsv_patch.sh

@@ -0,0 +1,45 @@
+#!/usr/bin/env bash
+set -uo pipefail ## Pipefail, complain on new unassigned variables.
+
+## Track the version of the TSV_patch template used
+VERSION='0.2.0dev'
+
+## This script is applied to the eager input TSV file locally to edit the dummy
+##    path to the fastQ files added by `create_eager_input.sh` to a real local
+##    path provided as a positional argument. Any further local tweaks to the
+##    TSV before running eager should be added below that in the form of bash
+##    commands to aid in reproducibility.
+
+## usage tsv_patch.sh <local_data_dir> <input_tsv>
+
+local_data_dir="$(readlink -f ${1})"
+input_tsv="$(readlink -f ${2})"
+output_tsv="$(dirname ${local_data_dir})/$(basename -s ".tsv" ${input_tsv}).finalised.tsv"
+columns_to_keep=("Sample_Name" "Library_ID" "Lane" "Colour_Chemistry" "SeqType" "Organism" "Strandedness" "UDG_Treatment" "R1" "R2" "BAM")
+source $(dirname ${2})/../../scripts/source_me.sh ## Load helper functions
+
+## Index non-proliferated columns and exclude them from the finalised TSV
+cut_selector=''
+tsv_header=($(head -n1 ${input_tsv}))
+for col_name in ${columns_to_keep[@]}; do
+  let idx=$(get_index_of ${col_name} "${columns_to_keep[@]}")+1 ## awk uses 1-based indexing
+  if [[ ! ${idx} -eq -1 ]]; then
+    cut_selector+="${idx},"
+  fi
+done
+
+## Remove added columns, and put columns in right order
+cut -f ${cut_selector%,} ${input_tsv} > ${output_tsv}
+sed -i -e "s|<PATH_TO_DATA>|${local_data_dir}|g" ${output_tsv}
+
+## Any further commands to edit the file before finalisation should be added below as shown
+# sed -ie 's/replace_this/with_this/g' ${output_tsv}
+
+## Keep track of versions
+version_file="$(dirname ${input_tsv})/script_versions.txt"
+##    Remove versions from older run if there
+grep -v -F -e "$(basename ${0})" -e "source_me.sh for final TSV" ${version_file} >${version_file}.new
+##    Then add new versions
+echo -e "$(basename ${0}):\t${VERSION}" >> ${version_file}.new
+echo -e "source_me.sh for final TSV:\t${HELPER_FUNCTION_VERSION}" >>${version_file}.new
+mv ${version_file}.new ${version_file}

packages/2023_Salazar_MachuPicchu/script_versions.txt

@@ -0,0 +1,2 @@
+create_eager_input.sh:	0.2.1dev
+source_me.sh for initial TSV:	0.2.1dev