Add local module duckdb/aminoacidhistogram

Author: heuermh

conf/base.config

@@ -59,10 +59,4 @@ process {
         errorStrategy = 'retry'
         maxRetries    = 2
     }
-    withName: SEQKIT_STATS {
-        ext.args = ' ' // turn off --all default argument
-    }
-    withName: DSHBIO_FASTATOPARQUET {
-        ext.args = '--alphabet protein'
-    }
 }

conf/modules.config

@@ -10,6 +10,8 @@
 ----------------------------------------------------------------------------------------
 */
 
+nextflow.enable.moduleBinaries = true
+
 process {
 
     publishDir = [
@@ -26,5 +28,10 @@ process {
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
-
+    withName: SEQKIT_STATS {
+        ext.args = ' ' // turn off --all default argument
+    }
+    withName: DSHBIO_FASTATOPARQUET {
+        ext.args = '--alphabet protein'
+    }
 }

modules/local/duckdb/aminoacidhistogram/environment.yml

@@ -0,0 +1,7 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/modules/environment-schema.json
+channels:
+  - conda-forge
+
+dependencies:
+  - conda-forge::duckdb-cli=1.0.0

modules/local/duckdb/aminoacidhistogram/main.nf

@@ -0,0 +1,30 @@
+process DUCKDB_AMINOACIDHISTOGRAM {
+    tag "${meta.id}"
+    label 'process_medium'
+
+    conda "${moduleDir}/environment.yml"    
+    container 'community.wave.seqera.io/library/duckdb-cli:1.0.0--a85d12a2a9de17c9'
+
+    input:
+    tuple val(meta), path(parquet)
+
+    output:
+    tuple val(meta), path("*.histogram.tsv"), emit: histogram
+    path "versions.yml"                     , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def sql = "INSTALL parquet; LOAD parquet; COPY (WITH p AS (SELECT * FROM read_parquet('${parquet}/*.parquet')), s AS (SELECT unnest(string_to_array(sequence, '')) AS aa FROM p), h AS (SELECT unnest(map_entries(histogram(aa))) AS kv FROM s), e AS (SELECT * from read_csv_auto('amino_acid_properties.tsv')) SELECT h.kv['key'] AS amino_acid, h.kv['value'] AS count, e.* FROM h JOIN e ON h.kv['key'] = e.one_letter_symbol) TO '${prefix}.histogram.tsv' (HEADER, DELIMITER '\t')"
+    """
+    create_amino_acid_properties.sh
+    duckdb :memory: "$sql"
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        duckdb: \$( duckdb --version | cut -f 1 -d " " )
+    END_VERSIONS
+    """
+}

modules/local/duckdb/aminoacidhistogram/resources/usr/bin/create_amino_acid_properties.sh

@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+#
+# See https://en.wikipedia.org/wiki/Amino_acid
+
+cat <<END_PROPERTIES > amino_acid_properties.tsv
+amino_acid	three_letter_symbol	one_letter_symbol	class	chemical_polarity	net_charge	hydropathy_index	molecular_mass	abundance_in_proteins	standard_genetic_coding	hydrophobic	aromatic	aliphatic	small	hydrophilic	positively_charged	negatively_charged
+Alanine	Ala	A	Aliphatic	Nonpolar	Neutral	1.8	89.094	8.76	GCN	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE	FALSE
+Arginine	Arg	R	Fixed cation	Basic polar	Positive	−4.5	174.203	5.78	MGR, CGY	FALSE	FALSE	FALSE	FALSE	TRUE	TRUE	FALSE
+Asparagine	Asn	N	Amide	Polar	Neutral	−3.5	132.119	3.93	AAY	FALSE	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE
+Aspartate	Asp	D	Anion	Brønsted base	Negative	−3.5	133.104	5.49	GAY	FALSE	FALSE	FALSE	FALSE	TRUE	FALSE	TRUE
+Cysteine	Cys	C	Thiol	Brønsted acid	Neutral	2.5	121.154	1.38	UGY	FALSE	FALSE	FALSE	FALSE	FALSE	FALSE	FALSE
+Glutamine	Gln	Q	Amide	Polar	Neutral	−3.5	146.146	3.9	CAR	FALSE	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE
+Glutamate	Glu	E	Anion	Brønsted base	Negative	−3.5	147.131	6.32	GAR	FALSE	FALSE	FALSE	FALSE	TRUE	FALSE	TRUE
+Glycine	Gly	G	Aliphatic	Nonpolar	Neutral	−0.4	75.067	7.03	GGN	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE	FALSE
+Histidine	His	H	Cationic	Brønsted acid and base	Positive, 10% Neutral, 90%	−3.2	155.156	2.26	CAY	FALSE	TRUE	FALSE	FALSE	TRUE	TRUE	FALSE
+Isoleucine	Ile	I	Aliphatic	Nonpolar	Neutral	4.5	131.175	5.49	AUH	TRUE	FALSE	TRUE	FALSE	FALSE	FALSE	FALSE
+Leucine	Leu	L	Aliphatic	Nonpolar	Neutral	3.8	131.175	9.68	YUR, CUY	TRUE	FALSE	TRUE	FALSE	FALSE	FALSE	FALSE
+Lysine	Lys	K	Cation	Brønsted acid	Positive	−3.9	146.189	5.19	AAR	FALSE	FALSE	FALSE	FALSE	TRUE	TRUE	FALSE
+Methionine	Met	M	Thioether	Nonpolar	Neutral	1.9	149.208	2.32	AUG	TRUE	FALSE	TRUE	FALSE	FALSE	FALSE	FALSE
+Phenylalanine	Phe	F	Aromatic	Nonpolar	Neutral	2.8	165.192	3.87	UUY	TRUE	TRUE	FALSE	FALSE	FALSE	FALSE	FALSE
+Proline	Pro	P	Cyclic	Nonpolar	Neutral	−1.6	115.132	5.02	CCN	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE	FALSE
+Serine	Ser	S	Hydroxylic	Polar	Neutral	−0.8	105.093	7.14	UCN, AGY	FALSE	FALSE	FALSE	TRUE	TRUE	FALSE	FALSE
+Threonine	Thr	T	Hydroxylic	Polar	Neutral	−0.7	119.119	5.53	ACN	FALSE	FALSE	FALSE	FALSE	TRUE	FALSE	FALSE
+Tryptophan	Trp	W	Aromatic	Nonpolar	Neutral	−0.9	204.228	1.25	UGG	TRUE	TRUE	FALSE	FALSE	FALSE	FALSE	FALSE
+Tyrosine	Tyr	Y	Aromatic	Brønsted acid	Neutral	−1.3	181.191	2.91	UAY	TRUE	TRUE	FALSE	FALSE	FALSE	FALSE	FALSE
+Valine	Val	V	Aliphatic	Nonpolar	Neutral	4.2	117.148	6.73	GUN	TRUE	FALSE	TRUE	FALSE	FALSE	FALSE	FALSE
+Selenocysteine	Sec	U					168.064									
+Pyrrolysine	Pyl	O					255.313									
+Any/unknown	Xaa	X														
+Asparagine or aspartate	Asx	B														
+Glutamine or glutamate	Glx	Z														
+Leucine or isoleucine	Xle	J														
+END_PROPERTIES

workflows/proteinannotator.nf

@@ -4,6 +4,7 @@
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 */
 include { DSHBIO_FASTATOPARQUET  } from '../modules/local/dshbio/fastatoparquet/main'
+include { DUCKDB_AMINOACIDHISTOGRAM  } from '../modules/local/duckdb/aminoacidhistogram/main'
 include { MULTIQC                } from '../modules/nf-core/multiqc/main'
 include { SEQKIT_STATS           } from '../modules/nf-core/seqkit/stats/main'
 include { paramsSummaryMap       } from 'plugin/nf-schema'
@@ -34,6 +35,10 @@ workflow PROTEINANNOTATOR {
     DSHBIO_FASTATOPARQUET(ch_samplesheet)
     ch_versions = ch_versions.mix(DSHBIO_FASTATOPARQUET.out.versions)
 
+    // todo: move this to stats on input fasta subworkflow
+    DUCKDB_AMINOACIDHISTOGRAM(DSHBIO_FASTATOPARQUET.out.parquet)
+    ch_versions = ch_versions.mix(DUCKDB_AMINOACIDHISTOGRAM.out.versions)
+
     //
     // Collate and save software versions
     //