feat: get refs from database (#69)

* feat: rule to download references

* fix: undefined output

* style. removed unnecessary dir creation

* fix: rules for obtaining reference data  work as intended now

* fix: genome to transcriptome gets input from ref.smk downloads instead of local directory

* test: added accesion no. to config to allow ref. data download

.test/config-simple/config.yml

@@ -13,6 +13,8 @@ repo: "https://github.com/snakemake-workflows/transriptome-differential-expressi
 
 ## Workflow-specific Parameters:
 
+# NCBI accession number
+accession: "GCA_917627325.4"
 # Genome fasta (absolute path)
 genome: "/lustre/miifs01/project/m2_zdvhpc/transcriptome_data/GCA_917627325.4_PGI_CHIRRI_v4_genomic.fa"
 # Annotation GFF/GTF (absolute path)

config/Mainz-MogonNHR/config.yml

@@ -13,6 +13,8 @@ repo: "https://github.com/snakemake-workflows/transriptome-differential-expressi
 
 ## Workflow-specific Parameters:
 
+# NCBI accession number
+accession: "GCA_917627325.4"
 # Genome fasta (absolute path)
 genome: "/lustre/miifs01/project/nhr-zdvhpc/transcriptome_data/GCA_917627325.4_PGI_CHIRRI_v4_genomic.fa"
 # Annotation GFF/GTF (absolute path)

workflow/Snakefile

@@ -14,6 +14,7 @@ configfile: "config/config.yml"
 include: "rules/commons.smk"
 include: "rules/qc.smk"
 include: "rules/utils.smk"
+include: "rules/ref.smk"
 include: "rules/datamod.smk"
 include: "rules/alignment.smk"
 include: "rules/alignmod.smk"

workflow/rules/datamod.smk

@@ -4,8 +4,8 @@ localrules:
 
 rule genome_to_transcriptome:
     input:
-        genome=config["genome"],
-        annotation=config["annotation"],
+        genome="references/genomic.fa",
+        annotation="references/genomic.gff",
     output:
         transcriptome="transcriptome/transcriptome.fa",
     log:

workflow/rules/ref.smk

@@ -0,0 +1,37 @@
+localrules:
+    get_genome,
+    get_annotation,
+
+
+rule get_genome:
+    output:
+        genome="references/genomic.fa",
+    params:
+        accession=config["accession"],
+    log:
+        "logs/refs/get_genome.log",
+    conda:
+        "../envs/env.yml"
+    shell:
+        """
+        curl -o data.zip https://api.ncbi.nlm.nih.gov/datasets/v2alpha/genome/accession/{params.accession}/download?include_annotation_type=GENOME_FASTA &> {log};
+        unzip -p data.zip ncbi_dataset/data/{params.accession}/*.fna > references/genomic.fa 2> {log};
+        rm data.zip &> {log}
+        """
+
+
+rule get_annotation:
+    output:
+        "references/genomic.gff",
+    params:
+        accession=config["accession"],
+    log:
+        "logs/refs/get_annotation.log",
+    conda:
+        "../envs/env.yml"
+    shell:
+        """
+        curl -o data.zip https://api.ncbi.nlm.nih.gov/datasets/v2alpha/genome/accession/{params.accession}/download?include_annotation_type=GENOME_GFF &> {log};
+        unzip -p data.zip ncbi_dataset/data/{params.accession}/*.gff > references/genomic.gff 2> {log};
+        rm data.zip &> {log}
+        """