moved from entrez to datasets

src/Afanc/autodatabase/runFuncs.py

@@ -12,7 +12,7 @@ def runAutoDB(args):
     """ Run autodatabase pipeline
     """
     from ..utilities.makeWD import initAutoDBDirStructure
-    from ..utilities.getVersions import get_versions_autodatabase
+    from ..utilities.getVersions import getVersionsAutodatabase
 
     subprocessID = "MAIN"
     vprint(
@@ -30,7 +30,7 @@ def runAutoDB(args):
     fasta_db_path = args.fastaDir
 
     ## capture python package and software versions for the autodatabase module in a JSON file
-    get_versions_autodatabase(args)
+    getVersionsAutodatabase(args)
 
     ## download ncbi taxonomy and preprocess fastas
     fasta_dict, mapping_dict = preprocessing(args, fasta_db_path)
@@ -42,13 +42,13 @@ def runAutoDB(args):
     makeK2db(args)
 
     ## make the variant index from quality controlled assemblies
-    make_variant_index(args)
+    makeVariantIndex(args)
 
     ## make a Krona chart for pleasing visualisation
     makeKronaChart(args)
 
     ## clean the output directory
-    clean_outdir(args)
+    cleanOutdir(args)
 
     vprint(
         "FINISHED",
@@ -205,7 +205,7 @@ def makeK2db(args):
     chdir(args.autoDB_WDir)
 
 
-def make_variant_index(args):
+def makeVariantIndex(args):
     """ Generates a variant index of parent child distances
     """
     from .makeVariantIndex import make_variant_index
@@ -251,7 +251,7 @@ def makeKronaChart(args):
     chdir(args.autoDB_WDir)
 
 
-def clean_outdir(args):
+def cleanOutdir(args):
     """ Cleans the output directory according to provided arguments.
 
     clean : remove the mash working directory.

src/Afanc/screen/runFuncs.py

@@ -24,7 +24,8 @@ def runScreen(args):
 
     ## if no_map is True then exit Afanc screen
     if args.no_map:
-        vprint("FINISHED", f"no_map mode finished. Metagenomic report can be found in {out_json}\n", "prGreen")
+        final_report = makeFinalReport(args, None, None)
+        vprint("FINISHED", f"no_map mode finished. Metagenomic report can be found in {final_report}\n", "prGreen")
         return 0
 
     ## parse kraken2 report to a json
@@ -44,7 +45,7 @@ def runScreen(args):
     final_report = makeFinalReport(args, variant_profile, reports)
 
     if args.clean or args.superclean:
-        final_report = clean_outdir(args, final_report)
+        final_report = cleanOutdir(args, final_report)
 
     vprint("FINISHED", f"Final report can be found at {final_report}\n", "prGreen")
 
@@ -193,7 +194,7 @@ def makeFinalReport(args, variant_profile, reports):
     import json
     from os import listdir
 
-    from Afanc.utilities.getVersions import get_versions_screen
+    from Afanc.utilities.getVersions import getVersionsScreen
 
     subprocessID = "REPORT"
     vprint(
@@ -228,7 +229,7 @@ def makeFinalReport(args, variant_profile, reports):
 
     ## collect versions
     jsondict["versions"] = {
-        "screen" : get_versions_screen()
+        "screen" : getVersionsScreen()
     }
 
     ## collect k2 json reports
@@ -243,25 +244,44 @@ def makeFinalReport(args, variant_profile, reports):
             ## initialise warnings
             event["warnings"] = []
 
-            ## block to deal with most likely variants
-            if "closest_variant" in event:
-                variant_flag = True
-                taxon_id = str(event["closest_variant"]["taxon_id"])
+            ## handle instances where an assembly cannot be found or no_map mode was used
+            if "assembly" in event:
 
-                ## in instances where this hit was subjected to variant profiling, the assembly used for mapping will
-                ## belong to the species rather than the closest variant
-                if "assembly" in event["closest_variant"]:
-                    assembly = event["closest_variant"]["assembly"]
+                ## block to deal with most likely variants
+                if "closest_variant" in event:
+                    variant_flag = True
+                    taxon_id = str(event["closest_variant"]["taxon_id"])
+
+                    ## in instances where this hit was subjected to variant profiling, the assembly used for mapping will
+                    ## belong to the species rather than the closest variant
+                    if "assembly" in event["closest_variant"]:
+                        assembly = event["closest_variant"]["assembly"]
+                    else:
+                        assembly = event["assembly"]
+
+                ## no variants
                 else:
+                    variant_flag = False
                     assembly = event["assembly"]
+                    taxon_id = str(event["taxon_id"])
+
+            ## block to deal with instances where there is no assembly
+            ## this is to ensure that a final json is constructed when no_map mode is used   
             else:
-                variant_flag = False
-                assembly = event["assembly"]
-                taxon_id = str(event["taxon_id"])
+                ## block to deal with most likely variants
+                if "closest_variant" in event:
+                    variant_flag = True
+                    assembly = None
+                    taxon_id = str(event["closest_variant"]["taxon_id"])
+
+                ## no variants
+                else:
+                    variant_flag = False
+                    assembly = None
+                    taxon_id = str(event["taxon_id"])
 
             ## block dealing with hits which have an accompanying assembly which reads were mapped to
             if not assembly == None:
-
                 assembly_prefix = path.basename(path.splitext(assembly)[0])
                 if assembly_prefix.endswith("_genomic"):
                     assembly_prefix = assembly_prefix.strip("_genomic")
@@ -310,7 +330,7 @@ def makeFinalReport(args, variant_profile, reports):
     return final_report
 
 
-def clean_outdir(args, final_report):
+def cleanOutdir(args, final_report):
     """ Cleans the output directory according to provided arguments.
 
     clean : remove the bt2 working directory.

src/Afanc/utilities/getVersions.py

@@ -5,7 +5,7 @@
 from .runCommands import command
 
 
-def get_versions_autodatabase(args):
+def getVersionsAutodatabase(args):
     """ get python package and software versions for afanc-autodatabase
     """
 
@@ -34,7 +34,7 @@ def get_versions_autodatabase(args):
         json.dump({ "afanc-autodatabase_versions" : version_dict }, fout, indent = 4)
 
 
-def get_versions_screen():
+def getVersionsScreen():
     """ Gets python package and softeware versions for afanc-screen
     """
 

test/afanc_test.py

@@ -24,7 +24,10 @@
 import argparse
 from argparse import RawTextHelpFormatter
 
-def run_test(args):
+
+def runTest(args):
+
+    nomap = True
 
     ## tsv for results
     tsv_handle = open("./Afanc_test.tsv", 'w')
@@ -61,16 +64,19 @@ def run_test(args):
             print(f"Error at {fq1} {fq2} {truth_val}. FASTQ files must end with either .fq.gz or .fastq.gz. Exiting.")
             sys.exit(1)
 
-        runline = f"afanc screen -o {prefix} -v {args.variants} -n 10 {args.db} {fq1} {fq2} -c > {prefix}.log"
-        # print(runline)
+        if nomap:
+            runline = f"afanc screen -o {prefix} -v {args.variants} -n 10 {args.db} {fq1} {fq2} -a > {prefix}.log"
+
+        else:
+            runline = f"afanc screen -o {prefix} -v {args.variants} -n 10 {args.db} {fq1} {fq2} -c > {prefix}.log"
 
         if not os.path.exists(prefix):
             subprocess.call(runline, shell=True)
 
         results_json = f"./{prefix}/{prefix}.json"
 
         if os.path.exists(results_json):
-            results = check_hits(results_json, truth_val)
+            results = checkHits(results_json, truth_val)
 
         else:
             result = "RUNFAIL"
@@ -94,24 +100,24 @@ def run_test(args):
 
     tsv_handle.close()
 
-def check_hits(results_json, truth_val):
+def checkHits(results_json, truth_val):
 
     with open(results_json, "r") as fin:
         jdata = json.load(fin)
 
         hits = jdata["results"]["Detection_events"]
 
-        cluster_results, cluster_hits = get_cluster_hits(hits["Clustering_results"], truth_val)
+        cluster_results, cluster_hits = getClusterHits(hits["Clustering_results"], truth_val)
 
         results = { "cluster" : [cluster_results, cluster_hits], "variant" : [None, [["None", "None"]]] }
 
         if "Variant_profile" in hits:
-            variant_results, variant_hits = get_variant_hits(hits["Variant_profile"], truth_val)
+            variant_results, variant_hits = getClusterHits(hits["Variant_profile"], truth_val)
             results["variant"] = [variant_results, variant_hits]
 
     return results
 
-def get_variant_hits(hits, truth_val):
+def getVariantHits(hits, truth_val):
 
     variant_box = []
 
@@ -130,7 +136,7 @@ def get_variant_hits(hits, truth_val):
 
     return result, variant_box
 
-def get_cluster_hits(hits, truth_val):
+def getClusterHits(hits, truth_val):
     """ checks hits against a known truth value
     """
 
@@ -184,6 +190,6 @@ def parse_args(argv):
     parser.add_argument("run_list", action='store', help="A tab separated txt file containing the paired end .fq paths and a truth value to compare the output to.")
     parser.add_argument("variants", action='store', help="A tab separated txt file containing variant info.")
 
-    run_test(parser.parse_args(argv))
+    runTest(parser.parse_args(argv))
 
 parse_args(sys.argv)