Contam emails (#171)

ChangeLog

@@ -1,6 +1,11 @@
 CHANGES
 =======
 
+* actual seq\_data dir in email
+* mycoplasma include in prep
+* include mycoplasma hyorhinis
+* auto version for docs, fix readthedocs yaml
+* update changelog
 * include authors
 * make sure doc pytest includes reqs from the doc folder
 * init readthedocs yaml, split up doc requirements

contaminome.yml

@@ -75,6 +75,11 @@ prokaryotes:
    vulgarname: mycoplasma
    accession: GCF_003385765.1
    taxid: 2148
+  Mesomycoplasma hyorhinis:
+    URL: https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/383/515/GCF_000383515.1_ASM38351v1/GCF_000383515.1_ASM38351v1_genomic.fna.gz
+    vulgarname: mycoplasma-hyorhinis
+    accession: GCF000383515.1
+    taxid: 2100
   Haemophilus influenzae:
    URL: https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/931/575/GCF_000931575.1_ASM93157v1/GCF_000931575.1_ASM93157v1_genomic.fna.gz
    vulgarname: haemophilus

src/tools/emailProjectFinished.py

@@ -5,6 +5,7 @@
 import os
 from dissectBCL.misc import getConf
 from email.mime.text import MIMEText
+import glob
 
 
 def fetchFirstNameAndEmail(lastName, config):
@@ -55,15 +56,30 @@ def fetchFirstNameAndEmail(lastName, config):
     return d[lastName]
 
 
-def getProjectIDs(projects):
+def getProjectIDs(projects, config):
     IDs = []
     for p in projects:
         # Sanity check
         assert (p.startswith("Project_"))
         IDs.append(p.split("_")[1])
+        PI = p.split("_")[-1].lower()
+    # Get the actual sequencing_data dir
+    # Assume if multiple projects are given, they all in the same flowcell.
+    flowcell = getFlowCell()
+    # Assume that only a flow cell exists only once.
+    seqdir = glob.glob(
+        os.path.join(
+            config['Dirs']['piDir'],
+            PI,
+            config['Internals']['seqDir'] + '*',
+            flowcell
+        )
+    )[0].split('/')[-2]
+
     if len(IDs) == 1:
-        return IDs[0]
-    return " and ".join([", ".join(IDs[:-1]), IDs[-1]])
+        return IDs[0], seqdir
+
+    return " and ".join([", ".join(IDs[:-1]), IDs[-1]]), seqdir
 
 
 def getFlowCell():
@@ -168,8 +184,11 @@ def main():
     if len(args.project) > 1:
         content += "s"
     content += (" {} are finished and the results are now available in your "
-                "group's sequencing data directory"
-                .format(getProjectIDs(args.project)))
+                "group's {} directory"
+                .format(
+                    getProjectIDs(args.project, config)[0],
+                    getProjectIDs(args.project, config)[1])
+                )
 
     content += " under the {} folder.\n".format(getFlowCell())
 

src/tools/prep_contaminome.py

@@ -52,6 +52,7 @@
     'pseudomonas-putidia': [1211579, 12, 'species'],
     'pseudomonas-fulva': [47880, 12, 'species'],
     'mycoplasma': [2148, 14, 'species'],
+    'mycoplasma-hyorhinis': [2100, 14, 'species'],
     'haemophilus': [727, 14, 'species'],
     'sars-cov2': [2697049, 7, 'species'],
     'flu-a': [335341, 7, 'species'],