From a835f79928bf6ec5c5b93678bd89bc54c59e3206 Mon Sep 17 00:00:00 2001
From: Thomas Weber <thomas.weber@embl.de>
Date: Tue, 17 May 2022 17:27:28 +0200
Subject: [PATCH] Download example & external data Implemented rules based on
snakemake.remote.HTTP function that can be called through config.yaml / CLI
arguments Update config.yaml file Update rules/examples.smk Update Snakefile
Update README.md
---
README.md | 2 +-
workflow/Snakefile | 48 ++++++++++++++++++++++++-------------
workflow/config/config.yaml | 12 +++++++---
workflow/rules/examples.smk | 34 ++++++++++++++------------
4 files changed, 61 insertions(+), 35 deletions(-)
diff --git a/README.md b/README.md
index d45f1fc..ac8e733 100644
--- a/README.md
+++ b/README.md
@@ -117,7 +117,7 @@ wget https://sandbox.zenodo.org/record/1062182/files/GCA_000001405.15_GRCh38_no_
#### 3A. Download example data automatically with snakemake [Optional]
```
-snakemake -c1 --config mode=download_example_data input_bam_location=/path/to/INPUT
+snakemake -c1 --config mode=download_data input_bam_location=/path/to/INPUT
```
**Warning:** Download example data currently requires 35GB of free space disk.
diff --git a/workflow/Snakefile b/workflow/Snakefile
index d51b840..fc024c8 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -13,12 +13,18 @@ from pprint import pprint
configfile: "config/config.yaml"
mode_selected = config["mode"].lower()
-correct_modes = ["setup", "count", "segmentation", "haplotagging", "strandphasing", "mosaiclassifier", "download_example_data"]
+correct_modes = ["count", "segmentation", "mosaiclassifier", "download_data"]
assert mode_selected in correct_modes, "Wrong mode selected : {}\nFollowing list of modes are available : {}".format(config["mode"], ", ".join(correct_modes))
plot_option_selected = config["plot"]
assert type(plot_option_selected) is bool, "Wrong plot option selected : {}\nPlease enter a valid value (True / False)".format(config["plot"])
+dl_bam_example_option_selected = config["dl_bam_example"]
+assert type(dl_bam_example_option_selected) is bool, "Wrong plot option selected : {}\nPlease enter a valid value (True / False)".format(config["plot"])
+
+dl_external_files_option_selected = config["dl_external_files"]
+assert type(dl_external_files_option_selected) is bool, "Wrong plot option selected : {}\nPlease enter a valid value (True / False)".format(config["plot"])
+
print(os.getcwd())
print("Mode selected : {}".format(mode_selected))
@@ -49,7 +55,7 @@ report: "report/workflow.rst"
if os.path.isfile(config["output_location"] + "config/config_df.tsv") is False:
-
+ print("HELLO")
######################################################################
# TODO : move to another file
@@ -79,10 +85,9 @@ if os.path.isfile(config["output_location"] + "config/config_df.tsv") is False:
# df["all/selected"] = df["Folder"].apply(lambda r: r.split("/")[-1])
-
+ print(config["input_bam_location"])
# Parsing folder and retrieve only files with .bam extension
data = [(r, file.replace(".bam", "")) for r, d, f in os.walk(config["input_bam_location"]) for file in f if ".bam" in file and ".bai" not in file]
-
# Building pandas df based on folder structure
df = pd.DataFrame(data, columns=["Folder", "File"])
@@ -196,14 +201,14 @@ if plot_option_selected == True:
# SETUP ENV
-if mode_selected == "setup":
- rule all:
- input:
- # rules.install_rlib_strandphaser.output,
- rules.compile_mosaic.output,
+# if mode_selected == "setup":
+# rule all:
+# input:
+# # rules.install_rlib_strandphaser.output,
+# rules.compile_mosaic.output,
# MODE MOSAIC COUNT
-elif mode_selected == "count":
+if mode_selected == "count":
if plot_option_selected == True:
rule all:
input:
@@ -291,9 +296,20 @@ elif mode_selected == "mosaiclassifier":
# # TEST MODE
-elif mode_selected == "download_example_data":
-
- rule all:
- input:
- rules.dl_example_data.output,
- rules.dl_external_data.output
+elif mode_selected == "download_data":
+ if dl_bam_example_option_selected is True and dl_external_files_option_selected is True:
+ rule all:
+ input:
+ rules.dl_example_data.output,
+ rules.dl_external_data.output
+
+ if dl_bam_example_option_selected is True and dl_external_files_option_selected is False:
+ rule all:
+ input:
+ rules.dl_example_data.output,
+
+ if dl_bam_example_option_selected is False and dl_external_files_option_selected is True:
+ rule all:
+ input:
+ rules.dl_external_data.output
+
diff --git a/workflow/config/config.yaml b/workflow/config/config.yaml
index e41a068..17606d6 100644
--- a/workflow/config/config.yaml
+++ b/workflow/config/config.yaml
@@ -12,18 +12,24 @@ mode: "count"
plot: False
### Enable / Disable comparison for each BAM file between folder name & SM tag
check_sm_tag: False
+### Enable / Disable download of BAM examples (RPE-BM510)
+dl_bam_example: False
+### Enable / Disable download of external files (1000G SNV & Fasta ref genome)
+dl_external_files: False
## Input BAM location
-input_bam_location: "TEST_EXAMPLE_DATA/bam/"
+input_bam_location: "TEST_EXAMPLE_DATA/RPE-BM510/"
## Output location
output_location: "TEST_OUTPUT/"
# External files
## 1000G SNV sites to genotype : https://sandbox.zenodo.org/record/1060653/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz
-snv_sites_to_genotype: "sandbox.zenodo.org/record/1060653/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz"
+snv_sites_to_genotype: "sandbox.zenodo.org/record/1062182/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz"
# Reference genome : https://sandbox.zenodo.org/record/1060653/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna
-reference: "sandbox.zenodo.org/record/1060653/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna"
+reference: "sandbox.zenodo.org/record/1062182/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna"
+
+tmp_check: "sandbox.zenodo.org/record/1062186/files/mosaic_logo.png"
# Chromosomes list to process
chromosomes:
diff --git a/workflow/rules/examples.smk b/workflow/rules/examples.smk
index 8555b07..6e27c81 100644
--- a/workflow/rules/examples.smk
+++ b/workflow/rules/examples.smk
@@ -4,27 +4,31 @@ from snakemake.remote.HTTP import RemoteProvider as HTTPRemoteProvider
HTTP = HTTPRemoteProvider()
rule dl_example_data:
+ """
+ rule fct: Download BAM example data as input for MosaiCatcher pipeline
+ input: zip file stored on Zenodo
+ output: input_bam_location given by the user
+ """
input:
- # HTTP.remote("https://git.embl.de/tweber/mosaicatcher-update/-/raw/dev/workflow/bam/RPE-BM510/all/BM510x04_PE20301.sort.mdup.bam", keep_local=True)
- HTTP.remote("https://sandbox.zenodo.org/record/1060987/files/TEST_EXAMPLE_DATA.zip", keep_local=True)
+ HTTP.remote("https://sandbox.zenodo.org/record/1062182/files/TEST_EXAMPLE_DATA.zip", keep_local=True)
output:
directory(config["input_bam_location"])
shell:
"mkdir {output};"
- "tar -xf {input} -C .;"
- "mv TEST_EXAMPLE_DATA/ {output}"
+ "unzip {input} -d .;"
+ "mv TEST_EXAMPLE_DATA/* {output}"
+# TODO: Adapt according reference
rule dl_external_data:
+ """
+ rule fct: Download External files
+ input: files stored on Zenodo
+ output: touch file to check if everything was running correctly
+ """
input:
- # HTTP.remote("https://git.embl.de/tweber/mosaicatcher-update/-/raw/dev/workflow/bam/RPE-BM510/all/BM510x04_PE20301.sort.mdup.bam", keep_local=True)
- HTTP.remote("https://sandbox.zenodo.org/record/1060987/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna", keep_local=True),
- HTTP.remote("https://sandbox.zenodo.org/record/1060987/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai", keep_local=True),
- HTTP.remote("https://sandbox.zenodo.org/record/1060987/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz", keep_local=True),
- HTTP.remote("https://sandbox.zenodo.org/record/1060987/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz.tbi", keep_local=True),
+ HTTP.remote("https://sandbox.zenodo.org/record/1062182/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna", keep_local=True),
+ HTTP.remote("https://sandbox.zenodo.org/record/1062182/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai", keep_local=True),
+ HTTP.remote("https://sandbox.zenodo.org/record/1062182/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz", keep_local=True),
+ HTTP.remote("https://sandbox.zenodo.org/record/1062182/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz.tbi", keep_local=True),
output:
- "sandbox.zenodo.org/record/1060987/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna",
- "sandbox.zenodo.org/record/1060987/files/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai",
- "sandbox.zenodo.org/record/1060987/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz"
- "sandbox.zenodo.org/record/1060987/files/ALL.chr1-22plusX_GRCh38_sites.20170504.renamedCHR.vcf.gz.tbi"
- shell:
- "echo Download completed"
+ touch(config["output_location"] + "config/dl_external_data.ok")
--
GitLab