handling sigs with empty names

.vscode/settings.json

@@ -1,6 +1,9 @@
 {
     "python.analysis.extraPaths": [
         "./src"
+    ],
+    "cSpell.words": [
+        "AMPLICON", "signame",
     ]
     // # show line numbers in jupyter notebooks
     // "jupyter.lineNumbers": "on",

src/snipe/api/multisig_reference_QC.py

@@ -391,12 +391,7 @@ def process_sample(self, sample_sig: SnipeSig, predict_extra_folds: Optional[Lis
 
 
         # ============= SAMPLE STATS =============
-
-        # if sample name is empty, we must set one with the file basename without extension
-        if not sample_sig.name:
-            sample_sig.name = os.path.basename(sample_sig.filename).split('.')[0]
-            self.logger.warning("Sample name is empty. Setting it to %s.", sample_sig.name)
-
+
         self.logger.debug("Processing sample statistics.")
         sample_stats_raw = sample_sig.get_sample_stats
         sample_stats.update({
@@ -734,9 +729,9 @@ def sort_chromosomes(chr_name):
 
             sample_nonref = sample_sig - self.reference_sig
 
-            self.logger.debug("\t-Size of non-reference k-mers in the sample signature: %d hashes.", len(sample_nonref))
+            self.logger.debug("\tSize of non-reference k-mers in the sample signature: %d hashes.", len(sample_nonref))
             # sample_nonref.trim_singletons()
-            self.logger.debug("\t-Size of non-reference k-mers after trimming singletons: %d hashes.", len(sample_nonref))
+            self.logger.debug("\tSize of non-reference k-mers after trimming singletons: %d hashes.", len(sample_nonref))
 
             sample_nonref_unique_hashes = len(sample_nonref)
             sample_nonref_total_abundance = sample_nonref.total_abundance
@@ -752,12 +747,11 @@ def sort_chromosomes(chr_name):
                 sample_nonref_var: SnipeSig = sample_nonref & variance_sig
 
                 if self.export_varsigs:
-                    __sample_name = sample_sig.name.replace(' ','_').lower()
-                    __var_name = variance_name.replace(' ','_').lower()
+                    __sample_name = sample_sig.name.replace(' ','_')
+                    __var_name = variance_name.replace(' ','_')
                     __filename = os.path.basename(f"{__sample_name}_{__var_name}_nonref.zip".strip())
                     self.logger.debug("Exporting non-reference k-mers from variable '%s'.", __filename)
-                    var_export_file_path = sample_nonref_var.export(__filename)
-                    sample_nonref_var.export(var_export_file_path)
+                    sample_nonref_var.export(__filename)
 
                 sample_nonref_var_total_abundance = sample_nonref_var.total_abundance
                 sample_nonref_var_fraction_total = sample_nonref_var_total_abundance / sample_nonref_total_abundance

src/snipe/api/snipe_sig.py

@@ -110,6 +110,13 @@ def __init__(self, *,
             else:
                 self.logger.debug("No signature found in the input. Expected a single sample signature.")
                 raise ValueError("No signature found in the input. Expected a single sample signature.")
+
+            if sig_type == SigType.AMPLICON:
+                for signame in sourmash_sigs.keys():
+                    if signame == '-' or signame == '':
+                        _emsg = "Amplicon signature must have a name!"
+                        self.logger.error(_emsg)
+                        raise ValueError(_emsg)
 
         elif sig_type == SigType.GENOME:
             if len(sourmash_sigs) > 1:
@@ -131,6 +138,10 @@ def __init__(self, *,
                         self.chr_to_sig[sig.name] = SnipeSig(sourmash_sig=sig, sig_type=SigType.AMPLICON, enable_logging=enable_logging)
                     elif signame.startswith("mitochondrial-"):
                         self.chr_to_sig[sig.name] = SnipeSig(sourmash_sig=sig, sig_type=SigType.AMPLICON, enable_logging=enable_logging)
+                    elif signame == '-' or signame == '':
+                        _emsg = "Reference signature must have a name!"
+                        self.logger.error(_emsg)
+                        raise ValueError(_emsg)
                     else:
                         continue
                 else:
@@ -143,14 +154,21 @@ def __init__(self, *,
         else:
             self.logger.debug("Unknown sigtype: %s", sig_type)
             raise ValueError(f"Unknown sigtype: {sig_type}")
-
+
+        _sourmash_sig_name = _sourmash_sig.name
+        if _sourmash_sig_name == '-':
+            _sourmash_sig_name = ''
+
         self.logger.debug("Length of currently loaded signature: %d, with name: %s", len(_sourmash_sig), _sourmash_sig.name)
+
+        self.logger.debug("Loaded %s with name %s", sig_type, _sourmash_sig_name)
 
         # Extract properties from the loaded signature
         self._ksize = _sourmash_sig.minhash.ksize
         self._scale = _sourmash_sig.minhash.scaled
         self._md5sum = _sourmash_sig.md5sum()
-        self._name = _sourmash_sig.name
+        self._name = _sourmash_sig_name
+        self.logger.debug("Loaded signature name is %s", self._name)
         self._filename = _sourmash_sig.filename
         self._track_abundance = _sourmash_sig.minhash.track_abundance
 
@@ -289,6 +307,7 @@ def name(self, name: str):
         r"""
         Set the name of the signature.
         """
+        self.logger.debug("Setting name to %s", name)
         self._name = name
 
     @track_abundance.setter
@@ -513,7 +532,7 @@ def _convert_to_sourmash_signature(self):
         self.sourmash_sig = sourmash.signature.SourmashSignature(mh, name=self._name, filename=self._filename)
         self.logger.debug("Conversion to sourmash.signature.SourmashSignature completed.")
 
-    def export(self, path, force=False) -> None:
+    def export(self, path) -> None:
         r"""
         Export the signature to a file.
 
@@ -528,9 +547,6 @@ def export(self, path, force=False) -> None:
                 sourmash.signature.save_signatures_to_json([self.sourmash_sig], fp)
 
         elif path.endswith(".zip"):
-            if os.path.exists(path): 
-                self.logger.debug(f"Output file already exists: {path}")
-                raise FileExistsError("Output file already exists.")
             try:
                 with sourmash.save_load.SaveSignatures_ZipFile(path) as save_sigs:
                     save_sigs.add(self.sourmash_sig)

src/snipe/cli/cli_qc.py

@@ -204,11 +204,11 @@ def process_subset(
     subset_failed = []
     for sample_path in subset:
         sample_sig = SnipeSig(sourmash_sig=sample_path, sig_type=SigType.SAMPLE, enable_logging=debug)
-        if len(sample_sig.name) == 0:
-            # warn and set to basename without extension
+        subset_logger.debug(f"DELME Processing sample: {sample_sig.name}")
+        if sample_sig.name == "":
             _newname = os.path.basename(sample_path).split('.')[0]
-            subset_logger.warning(f"Sample name is empty. Setting to: {_newname}")
-
+            sample_sig.name = _newname
+            subset_logger.warning(f"Sample name is empty. Setting to: `{sample_sig.name}`")
 
         try:
             sample_stats = qc_inst.process_sample(
@@ -678,6 +678,12 @@ def qc(ref: str, sample: List[str], samples_from_file: Optional[str],
         with tqdm(total=len(valid_samples), desc="Processing samples") as pbar:
             for sample_path in valid_samples:
                 sample_sig = SnipeSig(sourmash_sig=sample_path, sig_type=SigType.SAMPLE, enable_logging=debug)
+                qc_instance.logger.debug(f"DELME Processing sample: {sample_sig.name}")
+                if sample_sig.name == "":
+                    _newname = os.path.basename(sample_path).split('.')[0]
+                    sample_sig.name = _newname
+                    qc_instance.logger.warning(f"Sample name is empty. Setting to: `{sample_sig.name}`")
+
                 try:
                     sample_stats = qc_instance.process_sample(
                         sample_sig=sample_sig,
@@ -819,6 +825,12 @@ def qc(ref: str, sample: List[str], samples_from_file: Optional[str],
     metadata_str, metadata_md5sum = METADATA.export_and_verify_metadata(
         metadata=export_metadata
     )
+
+    # santize file_path and filename
+    df["filename"] = df["file_path"].apply(os.path.basename)
+    # drop file_path
+    df.drop(columns=["file_path"], inplace=True)
+
 
     try:
         with open(output, 'w', encoding='utf-8') as f: