FIX : Raise errors in the input files do not exist or if the organism…

… names contain spaces

VERSION

@@ -1 +1 @@
-1.0.4
+1.0.1

ppanggolin/annotate/annotate.py

@@ -284,7 +284,7 @@ def readAnnotations(pangenome, organisms_file, getSeq = True, pseudo = False):
     for line in bar:
         elements = [el.strip() for el in line.split("\t")]
         if len(elements)<=1:
-            logging.getLogger().error("No tabulation separator found in organisms file")
+            logging.getLogger().error(f"No tabulation separator found in given --fasta file: '{organisms_file}'")
             exit(1)
         bar.set_description("Processing "+elements[1].split("/")[-1])
         bar.refresh()

ppanggolin/cluster/cluster.py

@@ -142,7 +142,7 @@ def read_gene2fam(pangenome, gene2fam):
     link = True if pangenome.status["genomesAnnotated"] in ["Computed","Loaded"] else False
     if link:
         if len(gene2fam) != len(pangenome.genes):#then maybe there are genes with identical IDs
-            raise Exception("Something unexpected happened during clustering (have less genes clustered than genes in the pangenome). A probable reason is that two genes in two different organisms have the same IDs; If you are sure that all of your genes have an identical IDs, please post an issue at https://github.com/labgem/PPanGGOLiN/")
+            raise Exception("Something unexpected happened during clustering (have less genes clustered than genes in the pangenome). A probable reason is that two genes in two different organisms have the same IDs; If you are sure that all of your genes have non identical IDs, please post an issue at https://github.com/labgem/PPanGGOLiN/")
     bar = tqdm(gene2fam.items(), unit = "gene")
     for gene, (family, is_frag) in bar:
         fam = pangenome.addGeneFamily(family)
@@ -167,7 +167,7 @@ def writeGeneSequencesFromAnnotations(pangenome, fileObj):
     """
     logging.getLogger().info("Writing all of the CDS sequences for clustering...")
     bar =  tqdm(pangenome.genes, unit="gene")
-    for gene in bar:#reading the table chunk per chunk otherwise RAM dies on big pangenomes
+    for gene in bar:
         if gene.type == "CDS":
             fileObj.write('>' + gene.ID + "\n")
             fileObj.write(gene.dna + "\n")
@@ -210,8 +210,6 @@ def clustering(pangenome, tmpdir, cpu , defrag = False, code = "11", coverage =
 
     checkPangenomeForClustering(pangenome, tmpFile, force)
 
-
-
     logging.getLogger().info("Clustering all of the genes sequences...")
     rep, tsv = firstClustering(tmpFile, newtmpdir, cpu, code, coverage, identity)
     fam2seq = read_faa(rep)

ppanggolin/main.py

@@ -11,6 +11,7 @@
 import resource
 import pkg_resources
 import tempfile
+import os
 
 #local modules
 import ppanggolin.pangenome
@@ -24,11 +25,32 @@
 import ppanggolin.info
 import ppanggolin.align
 
-def requirements():
+def checkTsvSanity(tsv):
+    f = open(tsv,"r")
+    for line in f:
+        elements = [el.strip() for el in line.split("\t")]
+        if len(elements)<=1:
+            raise Exception(f"No tabulation separator found in given file: {tsv}")
+        if " " in elements[0]:
+            raise Exception(f"Your genome names contain spaces (The first encountered genome name that had this string : '{elements[0]}'). To ensure compatibility with all of the dependencies of PPanGGOLiN this is not allowed. Please remove spaces from your genome names.")
+
+def checkInputFiles(anno=None, pangenome=None, fasta=None):
     """
-        Checks if the tools and libraries required for each submodule are installed.
+        Checks if the provided input files exist and are of the proper format
     """
-    pass
+    if pangenome is not None:
+        if not os.path.exists(pangenome):
+            raise FileNotFoundError(f"No such file or directory: '{pangenome}'")
+
+    if anno is not None:
+        if not os.path.exists(anno):
+            raise FileNotFoundError(f"No such file or directory: '{anno}'")
+        checkTsvSanity(anno)
+
+    if fasta is not None:
+        if not os.path.exists(fasta):
+            raise FileNotFoundError(f"No such file or directory: '{fasta}'")
+        checkTsvSanity(fasta)
 
 def cmdLine():
 
@@ -95,6 +117,14 @@ def cmdLine():
 
 def main():
     args = cmdLine()
+
+    if hasattr(args, "pangenome"):
+        checkInputFiles(pangenome = args.pangenome)
+    if hasattr(args, "fasta"):
+        checkInputFiles(fasta = args.fasta)
+    if hasattr(args,"anno"):
+        checkInputFiles(anno = args.anno)
+
     if hasattr(args, "verbose"):
         if args.verbose == 2:
             level = logging.DEBUG#info, debug, warnings and errors