Merge pull request #5 from theiagen/smw-coverage-file-fix

Check if files exist and update the coverage value to be the correct one

Dockerfile

@@ -1,4 +1,4 @@
-ARG MERCURY_VER="1.0.8"
+ARG MERCURY_VER="1.0.9"
 
 FROM google/cloud-sdk:480.0.0-slim
 

README.md

@@ -27,16 +27,16 @@ Default databases by organism:
 We highly recommend using the following Docker image to run Mercury:
 
 ```bash
-docker pull us-docker.pkg.dev/general-theiagen/theiagen/mercury:1.0.8
+docker pull us-docker.pkg.dev/general-theiagen/theiagen/mercury:1.0.9
 ```
 
 The entrypoint for this Docker image is the Mercury help message. To run this container interactively, use the following command:
 
 ```bash
-docker run -it --entrypoint=/bin/bash us-docker.pkg.dev/general-theiagen/theiagen/mercury:1.0.8
+docker run -it --entrypoint=/bin/bash us-docker.pkg.dev/general-theiagen/theiagen/mercury:1.0.9
 # Once inside the container interactively, you can run the mercury tool
 python3 /mercury/mercury/mercury.py -v
-# v1.0.8
+# v1.0.9
 ```
 
 ### Locally with Python

mercury/Runner.py

@@ -43,7 +43,7 @@ def __init__(self, options):
     if self.clearlabs_data:
       self.read1_column_name = "clearlabs_fastq_gz"
       self.assembly_fasta_column_name = "clearlabs_fasta"
-      self.assembly_mean_coverage_column_name = "clearlabs_assembly_coverage"
+      self.assembly_mean_coverage_column_name = "clearlabs_sequencing_depth"
 
     if self.reads_dehosted:
       self.read1_column_name = "reads_dehosted"

mercury/Table.py

@@ -248,13 +248,21 @@ def make_sra_csv(self):
 
     self.logger.info("TABLE:Copying over SRA files to the indicated GCP bucket ({})".format(self.gcp_bucket_uri))
     for oldname, newname in read_tuples:
-      transfer_command = "gcloud storage cp " + oldname + " " + self.gcp_bucket_uri + "/" + newname
-      self.logger.debug("TABLE:Running command: " + transfer_command)
+      check_if_transferred_command = "gcloud storage ls " + self.gcp_bucket_uri + "/" + newname
+      self.logger.debug("TABLE:Running command: " + check_if_transferred_command)
       try:
-        subprocess.run(transfer_command, shell=True, check=True)
+        subprocess.run(check_if_transferred_command, shell=True, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+        self.logger.warning("TABLE:Warning: A file with an identical name was found in the destination Google bucket; data transfer will be skipped.")
       except:
-        self.logger.error("TABLE:Error: non-zero exit code when copying files to GCP bucket ({})".format(self.gcp_bucket_uri))
-        sys.exit(1)
+        self.logger.debug("TABLE:The data did not appear in the bucket; now attempting to transfer the data")
+
+        transfer_command = "gcloud storage cp " + oldname + " " + self.gcp_bucket_uri + "/" + newname
+        self.logger.debug("TABLE:Running command: " + transfer_command)
+        try:
+          subprocess.run(transfer_command, shell=True, check=True)
+        except:
+          self.logger.error("TABLE:Error: non-zero exit code when copying files to GCP bucket ({})".format(self.gcp_bucket_uri))
+          sys.exit(1)
 
     self.logger.info("TABLE:Files copied to the indicated GCP bucket ({})".format(self.gcp_bucket_uri))
     self.logger.debug("TABLE:SRA metadata file created and data transferred")

mercury/__init__.py

@@ -1 +1 @@
-__VERSION__ = "v1.0.8"
+__VERSION__ = "v1.0.9"