More comprehensive example data

ctglab · Sep 3, 2024 · 3f33407 · 3f33407
1 parent 8815c18
commit 3f33407
Show file tree

Hide file tree

Showing 20 changed files with 12 additions and 8 deletions.
diff --git a/config/config_main.yaml b/config/config_main.yaml
@@ -54,13 +54,13 @@ datadirs:
 
 resources:
   # -- references -- #
-  genome: "test_data/genome_chr6.fa"
+  genome: "test_data/genome_chr6.fa.gz"
   transcriptome: "test_data/chr6_cdna.fa.gz"
   gtf: "test_data/chr6_105.gtf"
   # -- vcfs -- #
   gnomad: "test_data/gnomad_chr6.vcf.gz"
   gsnps: "test_data/1000G_snsp_chr6.vcf.gz"
-  dbsnps: "test_data/dbsnpALFA_chr6_filt.vcf.gz"
+  dbsnps: "test_data/dbsnpALFA_chr6.vcf.gz"
   REDI: "test_data/REDI_chr6.BED.gz"
   small_exac: "test_data/exac_chr6.vcf.gz"
   indel: "test_data/indels_chr6.vcf.gz"

diff --git a/test_data/1000G_snsp_chr6.vcf.gz b/test_data/1000G_snsp_chr6.vcf.gz
diff --git a/test_data/1000G_snsp_chr6.vcf.gz.tbi b/test_data/1000G_snsp_chr6.vcf.gz.tbi
diff --git a/test_data/chr6_cdna.fa.gz b/test_data/chr6_cdna.fa.gz
diff --git a/test_data/dbsnpALFA_chr6_filt.vcf.gz → test_data/dbsnpALFA_chr6.vcf.gz b/test_data/dbsnpALFA_chr6_filt.vcf.gz → test_data/dbsnpALFA_chr6.vcf.gz
diff --git a/test_data/dbsnpALFA_chr6.vcf.gz.tbi b/test_data/dbsnpALFA_chr6.vcf.gz.tbi
diff --git a/test_data/dbsnpALFA_chr6_filt.vcf.gz.tbi b/test_data/dbsnpALFA_chr6_filt.vcf.gz.tbi
diff --git a/test_data/exac_chr6.vcf.gz b/test_data/exac_chr6.vcf.gz
diff --git a/test_data/exac_chr6.vcf.gz.tbi b/test_data/exac_chr6.vcf.gz.tbi
diff --git a/test_data/genome_chr6.dict b/test_data/genome_chr6.dict
@@ -1,2 +1,2 @@
-@HD	VN:1.0	SO:unsorted
-@SQ	SN:6	LN:31500000	M5:148749b4e34a789ec672d5af905cc509	UR:file:///g100_work/IscrC_ENEO2/repos/RNA-neoflow/test_data/genome_chr6.fa
+@HD	VN:1.6
+@SQ	SN:6	LN:170805979	M5:6a48dfa97e854e3c6f186c8ff973f7dd	UR:file:/Volumes/HD2/home/danilo/ENEO/test_data/genome_chr6.fa.gz
diff --git a/test_data/genome_chr6.fa.fai b/test_data/genome_chr6.fa.fai
@@ -1 +1 @@
-6	31500000	55	60	61
+6	170805979	56	60	61
diff --git a/test_data/genome_chr6.fa → test_data/genome_chr6.fa.gz b/test_data/genome_chr6.fa → test_data/genome_chr6.fa.gz
diff --git a/test_data/genome_chr6.fa.gz.fai b/test_data/genome_chr6.fa.gz.fai
@@ -0,0 +1 @@
+6	170805979	56	60	61
diff --git a/test_data/genome_chr6.fa.gz.gzi b/test_data/genome_chr6.fa.gz.gzi
diff --git a/test_data/gnomad_chr6.vcf.gz b/test_data/gnomad_chr6.vcf.gz
diff --git a/test_data/gnomad_chr6.vcf.gz.tbi b/test_data/gnomad_chr6.vcf.gz.tbi
diff --git a/test_data/indels_chr6.vcf.gz b/test_data/indels_chr6.vcf.gz
diff --git a/test_data/indels_chr6.vcf.gz.tbi b/test_data/indels_chr6.vcf.gz.tbi
diff --git a/workflow/rules/common.smk b/workflow/rules/common.smk
@@ -25,7 +25,7 @@ logpath.mkdir(parents=True, exist_ok=True)
 germProb_script = config["resources"]["germline_prob_script"]
 bam_final_path = config["datadirs"]["BQSR"]
 ref_fasta = config["resources"]["genome"]
-ref_dict = ref_fasta.replace(".fa", ".dict")
+ref_dict = ref_fasta.replace(".fa.gz", ".dict")
 intervals_path = os.path.join(
     config["OUTPUT_FOLDER"] + config["datadirs"]["utils"], "interval-files"
 )

diff --git a/workflow/rules/index.smk b/workflow/rules/index.smk
@@ -20,8 +20,11 @@ rule star_index:
         ncpus=8,
         time="6:00:00",
     shell:
-        """STAR --runMode genomeGenerate --runThreadN {threads} --genomeDir {output} \
-        --genomeFastaFiles {input.fasta} --sjdbOverhang 100 --sjdbGTFfile {input.gtf}"""
+        """
+        mkfifo genome
+        zcat {input.fasta} > genome &
+        STAR --runMode genomeGenerate --runThreadN {threads} --genomeDir {output} \
+        --genomeFastaFiles genome --sjdbOverhang 100 --sjdbGTFfile {input.gtf}"""
 
 
 rule salmon_gentrome: