Merge pull request #169 from polio-nanopore/paf-parsing

yaml parsing and dev info and citation box in report

piranha/command.py

@@ -74,7 +74,7 @@ def main(sysargs = sys.argv[1:]):
     o_group.add_argument('--save-config',action="store_true",help=f"Output the config file with all parameters used")
 
     misc_group = parser.add_argument_group('Misc options')
-    misc_group.add_argument('--runname',action="store",help=f"Run name to appear in report. Default: {VALUE_RUN_NAME}")
+    misc_group.add_argument('--runname',action="store",help=f"Run name to appear in report. Default: {VALUE_RUNNAME}")
     misc_group.add_argument('--username',action="store",help="Username to appear in report. Default: no user name")
     misc_group.add_argument('--institute',action="store",help="Institute name to appear in report. Default: no institute name")
     misc_group.add_argument('--orientation',action="store",help="Orientation of barcodes in wells on a 96-well plate. If `well` is supplied as a column in the barcode.csv, this default orientation will be overwritten. Default: `vertical`. Options: `vertical` or `horizontal`.")

piranha/data/report.mako

@@ -58,6 +58,15 @@
           padding: 10px;
           flex: 50%;
         }
+      .info_box {
+          background-color: #eee;
+          color: #444;
+          padding: 13px;
+          width: 100%;
+          border: none;
+          text-align: left;
+          outline: none;
+        }
       .accordion {
           background-color: #eee;
           color: #444;
@@ -581,24 +590,28 @@
       <a href="#" id="toTopBtn" class="cd-top text-replace js-cd-top cd-top--is-visible cd-top--fade-out" data-abc="true"></a>
       <div>
       <header class="piranha-header">
-        <div class="col-sm-4" style="text-align: left;">
-          <img class="piranha-logo" src="https://raw.githubusercontent.com/aineniamh/piranha/main/docs/poseco.svg" vertical-align="left" width="30" height="30"></img>
+        <div class="col-sm-4" style="text-align: left;"><a href="https://polionanopore.org/">
+          <img class="piranha-logo" src="https://raw.githubusercontent.com/aineniamh/piranha/main/docs/poseco.svg" vertical-align="left" width="30" height="30"></img></a>
           PoSeCo | <small class="text-muted">${LANGUAGE_CONFIG["2"]}</small>
         </div>
-        <div class="col-sm-8" style="text-align: right;">
-          piranha | <small class="text-muted">${LANGUAGE_CONFIG["3"]}</small>
+        <div class="col-sm-8" style="text-align: right;"><a href="https://github.com/polio-nanopore/piranha/">
+          <strong style="color:#000"> piranha | </strong> <small class="text-muted">${LANGUAGE_CONFIG["3"]}</small></a>
         </div>
         <br>
         <hr>
       </header>
 
       <h1>${LANGUAGE_CONFIG["4"]} ${run_name} <small class="text-muted" style="color:${themeColor}">${date}</small></h1>
+      <br>
       %if config["username"]!="":
         <h3><strong>${LANGUAGE_CONFIG["5"]}</strong> | ${config["username"].lstrip("'").rstrip("'")}</h3>
       %endif
       %if config["institute"]!="":
         <h3><strong>${LANGUAGE_CONFIG["6"]}</strong> | ${config["institute"].lstrip("'").rstrip("'")}</h3>
       %endif
+      <div class="info_box">
+        <p>${LANGUAGE_CONFIG["52"]}</p>
+      </div>
       <br>
       %if flagged_high_npev and config["sample_type"]=="environmental":
       <h3><strong>${LANGUAGE_CONFIG["7"]}</strong> | ${LANGUAGE_CONFIG["8"]}</h3>
@@ -1007,6 +1020,12 @@
               });
             }
       </script>
+    <div id="citation_box" class="info_box">
+      <p>${LANGUAGE_CONFIG["53"]}</p>
+      <p>
+        <strong>O’Toole Á, Colquhoun R, Ansley C, Troman C, Maloney D, Vance Z, Akello J, Bujaki E, Majumdar M, Khurshid A, Arshad Y, Alam MM, Martin J, Shaw A, Grassly N, Rambaut A</strong> (2023) Automated detection and classification of polioviruses from nanopore sequencing reads using piranha. <i>bioRxiv</i> <a style='color:#e68781' href="https://doi.org/10.1101/2023.09.05.556319">https://doi.org/10.1101/2023.09.05.556319</a>
+      </p>
+    </div>
     <footer class="page-footer">
       <div class="container-fluid text-right text-md-right">
         <hr>

piranha/input_parsing/initialising.py

@@ -53,7 +53,7 @@ def get_defaults():
                     KEY_RUN_PHYLO:False,
                     KEY_CLUSTERS:[],
                     KEY_ANNOTATIONS:"",
-                    KEY_TREE_ANNOTATIONS:"source ",
+                    KEY_TREE_ANNOTATIONS:VALUE_TREE_ANNOTATIONS,
                     KEY_SAMPLE_SEQS:"",
                     KEY_PHYLO_METADATA_COLUMNS:VALUE_PHYLO_METADATA_COLUMNS,
                     KEY_SUPPLEMENTARY_METADATA_COLUMNS:VALUE_SUPPLEMENTARY_METADATA_COLUMNS,
@@ -62,7 +62,7 @@ def get_defaults():
                     # misc defaults
                     KEY_ORIENTATION:VALUE_ORIENTATION,
                     KEY_LANGUAGE:VALUE_LANGUAGE,
-                    KEY_RUN_NAME:VALUE_RUN_NAME,
+                    KEY_RUNNAME:VALUE_RUNNAME,
                     KEY_USERNAME:"",
                     KEY_POSITIVE:VALUE_POSITIVE,
                     KEY_NEGATIVE:VALUE_NEGATIVE,
@@ -81,19 +81,53 @@ def valid_args():
     return [
         KEY_READDIR,
         KEY_BARCODES_CSV,
+
         KEY_OUTDIR,
         KEY_OUTPUT_PREFIX,
+        KEY_PUBLISHDIR,
         KEY_TEMPDIR,
         KEY_NO_TEMP,
+        KEY_DATESTAMP,
         KEY_OVERWRITE,
+
+        KEY_SAMPLE_TYPE,
+        KEY_ANALYSIS_MODE,
+
         KEY_MIN_READ_LENGTH,
         KEY_MAX_READ_LENGTH,
         KEY_MIN_READS,
         KEY_MIN_PCENT,
+        KEY_MIN_MAP_QUALITY,
+        KEY_MIN_ALN_BLOCK,
+        KEY_PRIMER_LENGTH,
+
+        KEY_OUTPUT_REPORT,
+
+        KEY_REFERENCE_SEQUENCES,
+        KEY_REFERENCES_FOR_CNS,
+        KEY_MEDAKA_MODEL,
+
+        KEY_RUN_PHYLO,
+        KEY_SUPPLEMENTARY_SEQUENCES,
+        KEY_SUPPLEMENTARY_METADATA,
+        KEY_SUPPLEMENTARY_METADATA_ID_COLUMN,
+        KEY_TREE_ANNOTATIONS,
+        KEY_SUPPLEMENTARY_METADATA_COLUMNS,
+
+        KEY_POSITIVE,
+        KEY_NEGATIVE,
+        KEY_ORIENTATION,
+        KEY_DETAILED_TABLE_HEADER,
+
+        KEY_USERNAME,
+        KEY_INSTITUTE,
+        KEY_RUNNAME,
+
         KEY_THREADS,
         KEY_VERBOSE,
-        KEY_REFERENCE_SEQUENCES,
-        KEY_REFERENCES_FOR_CNS
+        KEY_COLOUR_MAP,
+        KEY_COLOUR_THEME,
+        KEY_LANGUAGE
     ]
 
 
@@ -137,7 +171,15 @@ def load_yaml(f):
     return input_config
 
 def return_path_keys():
-    return [KEY_READDIR,KEY_OUTDIR,KEY_TEMPDIR]
+    return [KEY_READDIR,
+            KEY_OUTDIR,
+            KEY_TEMPDIR,
+            KEY_PUBLISHDIR,
+            KEY_BARCODES_CSV,
+            KEY_REFERENCE_SEQUENCES,
+            KEY_SUPPLEMENTARY_SEQUENCES,
+            KEY_SUPPLEMENTARY_METADATA
+            ]
 
 def setup_absolute_paths(path_to_file,value):
     return os.path.join(path_to_file,value)
@@ -149,7 +191,7 @@ def parse_yaml_file(configfile,configdict):
 
     path_to_file = os.path.abspath(os.path.dirname(configfile))
     configdict[KEY_INPUT_PATH] = path_to_file
-    valid_keys = valid_args()
+    valid_keys = [k for k in globals().values() if type(k)==str]
     ignore_keys = ignore_args(configdict, valid_keys)
 
     invalid_keys = []
@@ -165,9 +207,10 @@ def parse_yaml_file(configfile,configdict):
                 clean_key = key.lstrip("-").replace("-","_").rstrip(" ").lstrip(" ").lower()
 
                 if clean_key not in valid_keys:
-                    if clean_key in ignore_keys:
-                        ignored_keys.append(key)
-                        continue
+                    print(valid_keys)
+                    # if clean_key in ignore_keys:
+                        # ignored_keys.append(key)
+                        # continue
                     invalid_keys.append(key)
                     break
 
@@ -212,7 +255,7 @@ def misc_args_to_config(verbose,threads,username,institute,run_name,config):
     misc.add_arg_to_config(KEY_THREADS,threads,config)
     misc.add_arg_to_config(KEY_USERNAME,username,config)
     misc.add_arg_to_config(KEY_INSTITUTE,institute,config)
-    misc.add_arg_to_config(KEY_RUN_NAME,run_name,config)
+    misc.add_arg_to_config(KEY_RUNNAME,run_name,config)
 
 def set_up_verbosity(config):
     if config[KEY_VERBOSE]:

piranha/report/make_report.py

@@ -595,7 +595,7 @@ def make_output_report(report_to_generate,barcodes_csv,preprocessing_summary,sam
 
     ctx = Context(buf, 
                     date = date.today(),
-                    run_name=config[KEY_RUN_NAME],
+                    run_name=config[KEY_RUNNAME],
                     version = __version__,
                     show_control_table = show_control_table,
                     plate_json = plate_json,

piranha/utils/config.py

@@ -95,7 +95,7 @@
 # MISC KEYS
 KEY_USERNAME="username"
 KEY_INSTITUTE="institute"
-KEY_RUN_NAME="run_name"
+KEY_RUNNAME="runname"
 KEY_VERBOSE="verbose"
 KEY_THREADS="threads"
 KEY_LOG_API="log_api"
@@ -164,10 +164,13 @@
 # report defaults
 VALUE_ORIENTATION="vertical"
 VALID_ORIENTATION=["vertical","horizontal"]
-VALUE_RUN_NAME="Nanopore"
+VALUE_RUNNAME="Nanopore"
 VALUE_COLOUR_MAP=["#e68781","#476970","#fbedac"]
 VALUE_COLOUR_THEME="#e68781"
 
+# phylo defaults
+VALUE_TREE_ANNOTATIONS="source "
+
 #file headers
 VARIANT_CALLS_HEADER_FIELDS = ["barcode","reference","variant_count","variants"]
 SAMPLE_SUMMARY_TABLE_HEADER_FIELDS = ["sample","barcode","Sample classification","reference_group","Number of mutations"]

piranha/utils/report_config.py

@@ -54,7 +54,9 @@
     "48" : "Export image",
     "49" : "snipit plot for queries in",
     "50" : "Co-occurrence matrix of Reference and Variant alleles called against",
-    "51" : ". This is the percentage of bases that cover those sites in the mapping file that are of a high quality (>13) and that are either the reference allele or the allele of the variant called at that site."
+    "51" : ". This is the percentage of bases that cover those sites in the mapping file that are of a high quality (>13) and that are either the reference allele or the allele of the variant called at that site.",
+    "52" : "This report is generated by <strong>piranha</strong>, a software tool developed to help standardise and streamline the analysis of poliovirus sequencing reads. It has been developed by members of the Rambaut group at the University of Edinburgh as part of the Poliovirus Sequencing Consortium. If you have used piranha in your work, please <a href='#citation_box' style='color:#e68781'><strong>cite</strong></a> the software.",
+    "53" : "General citation for Piranha"
 }
 
 # FRENCH REPORT
@@ -112,5 +114,7 @@
     "48" : "Exporter une image",
     "49" : "Graphique d'extraits pour les requêtes dans",
     "50" : "Matrice de cooccurrence des allèles de référence et de variante appelée par rapport à la référence",
-    "51" : ". Il s'agit du pourcentage de bases qui couvrent les sites du fichier de cartographie qui sont de haute qualité (>13) et qui sont soit l'allèle de référence, soit l'allèle du variant appelé sur ce site."
- }
+    "51" : ". Il s'agit du pourcentage de bases qui couvrent les sites du fichier de cartographie qui sont de haute qualité (>13) et qui sont soit l'allèle de référence, soit l'allèle du variant appelé sur ce site.",
+    "52" : "Ce rapport est généré par piranha, un outil logiciel développé pour aider à standardiser et rationaliser l'analyse des lectures de séquençage du poliovirus. Il a été développé par des membres du groupe Rambaut de l'Université d'Édimbourg dans le cadre du Poliovirus Sequencing Consortium. Si vous avez utilisé piranha dans votre travail, merci de <a href='#citation_box' style='color:#e68781'><strong>citer</strong></a> le logiciel.",
+    "53" : "Citation générale pour Piranha"
+    }