FunVIP 0.3.24.5__tar.gz → 0.3.24.7__tar.gz

{funvip-0.3.24.5 → funvip-0.3.24.7}/FunVIP.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: FunVIP
-Version: 0.3.24.5
+Version: 0.3.24.7
 Summary: Fungal Validation & Identification Pipeline
 Author-email: Changwan Seo <wan101010@snu.ac.kr>
 License: GPL-3.0

{funvip-0.3.24.5 → funvip-0.3.24.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: FunVIP
-Version: 0.3.24.5
+Version: 0.3.24.7
 Summary: Fungal Validation & Identification Pipeline
 Author-email: Changwan Seo <wan101010@snu.ac.kr>
 License: GPL-3.0

{funvip-0.3.24.5 → funvip-0.3.24.7}/funvip/main.py

@@ -34,6 +34,9 @@ def main():
     import logging
     import PyQt5
 
+    # To prevent pyqt error
+    os.environ["QT_QPA_PLATFORM"] = "offscreen"
+
     # To initialize logger to print ANSI color
     _ = subprocess.call("", shell=True)
 

{funvip-0.3.24.5 → funvip-0.3.24.7}/funvip/src/dataset.py

@@ -206,6 +206,9 @@ class FunVIP_var:
 
         for group in self.list_group:
             logging.info(f"Generating dataset for {group}")
+
+            print(f"opt.queryonly: {opt.queryonly}")
+
             dict_funinfo[group] = {}
 
             # For queryonly case
@@ -277,28 +280,36 @@ class FunVIP_var:
                     ]
                     self.add_dataset(group, "concatenated", list_qr, list_db, [])
 
-            # For opt.queryonly is False -> run all dataset in database
+            # For opt.queryonly is False -> run all dataset in database if possible
             else:
                 for gene in self.list_db_gene:
-                    list_qr = [
-                        FI
-                        for FI in self.list_FI
+                    list_qr = []
+                    for FI in self.list_FI:
                         if (
-                            gene in FI.seq
-                            and FI.datatype == "query"
+                            FI.datatype == "query"
                             and FI.adjusted_group == group
-                        )
-                    ]
-                    list_db = [
-                        FI
-                        for FI in self.list_FI
+                            and gene in FI.seq
+                        ):
+                            if FI.seq[gene] != "":
+                                list_db.append(FI)
+
+                    list_db = []
+                    for FI in self.list_FI:
                         if (
-                            gene in FI.seq
-                            and FI.datatype == "db"
+                            FI.datatype == "db"
                             and FI.adjusted_group == group
+                            and gene in FI.seq
+                        ):
+                            if FI.seq[gene] != "":
+                                list_db.append(FI)
+
+                    # If none of the database is possible for this group and gene pair, it should be excluded
+                    if len(list_db) > 0:
+                        self.add_dataset(group, gene, list_qr, list_db, [])
+                    else:
+                        logging.warning(
+                            f"dataset {group} {gene} did not passed dataset construction due to lack of sequences"
                         )
-                    ]
-                    self.add_dataset(group, gene, list_qr, list_db, [])
 
                 # for concatenated
                 list_qr = [
@@ -685,7 +696,7 @@ class FunVIP_var:
 
         # Add issue: the number of sequences are insufficient
         for fail in final_fail_list:
-            for FI in V.list_FI:
+            for FI in self.list_FI:
                 if FI.adjusted_group == fail[0]:
                     if FI.seq[fail[1]] != "":
                         FI.issues.add(f"lackseq")

{funvip-0.3.24.5 → funvip-0.3.24.7}/funvip/src/ext.py

@@ -292,10 +292,10 @@ def ModelFinder(fasta, opt, path, thread):
     if platform == "win32":
         if " " in fasta:
             fasta = f'"{fasta}"'
-        CMD = f"{path.sys_path}/external/iqtree/bin/iqtree2.exe --seqtype DNA -s {fasta} {model_term} -merit {opt.criterion} -T {thread} -mem {opt.memory}"
+        CMD = f"{path.sys_path}/external/iqtree/bin/iqtree2.exe --seqtype DNA -s {fasta} {model_term} -merit {opt.criterion} -nt AUTO -ntmax {thread} -mem {opt.memory} --quiet"
     else:
         # not final
-        CMD = f"iqtree --seqtype DNA -s '{fasta}' {model_term} -merit {opt.criterion} -T {thread} -mem {opt.memory}"
+        CMD = f"iqtree --seqtype DNA -s '{fasta}' {model_term} -merit {opt.criterion} -nt AUTO -ntmax {thread} -mem {opt.memory} --quiet"
     logging.info(CMD)
     Run = subprocess.call(CMD, shell=True)
 
@@ -418,9 +418,9 @@ def IQTREE(
         else:
             tmp_fasta = fasta
 
-        CMD = f"{path.sys_path}/external/iqtree/bin/iqtree2.exe -s {tmp_fasta} -B {bootstrap} -T {thread} {model}"
+        CMD = f"{path.sys_path}/external/iqtree/bin/iqtree2.exe -s {tmp_fasta} -B {bootstrap} -nt AUTO -ntmax {thread} {model}"
     else:
-        CMD = f"iqtree -s {fasta} -B {bootstrap} -T {thread} {model}"
+        CMD = f"iqtree -s {fasta} -B {bootstrap} -nt AUTO -ntmax {thread} {model}"
 
     logging.info(f"partition: {partition}")
     # Partitioned analysis cannot be used with memory option

{funvip-0.3.24.5 → funvip-0.3.24.7}/funvip/src/modeltest.py

@@ -1,6 +1,9 @@
 from funvip.src import ext
-import os, shutil, copy
+import os
+import shutil
+import copy
 import logging
+import multiprocessing as mp
 
 cmd_translator_modeltestng_fasttree = {
     "JC": "",
@@ -362,9 +365,9 @@ def modeltest(V, path, opt) -> dict:
     model_dict = copy.deepcopy(group_dict)
 
     # By group and by gene
-    for group in group_dict:
-        for gene in group_dict[group]:
-            if opt.method.modeltest.lower() == "modeltest-ng":
+    if opt.method.modeltest.lower() == "modeltest-ng":
+        for group in group_dict:
+            for gene in group_dict[group]:
                 # As modeltest-ng shows only top 10 results,
                 # we should reduce number of models to prevent none of the model fits
                 if opt.method.tree == "fasttree":
@@ -397,17 +400,41 @@ def modeltest(V, path, opt) -> dict:
                     path=path,
                     opt=opt,
                 )
-            elif opt.method.modeltest.lower() == "iqtree":
-                if not (opt.method.tree == "iqtree"):
-                    # Run IQTREE ModelFinder
-                    ext.ModelFinder(
-                        fasta=f"{path.out_alignment}/{opt.runname}_trimmed_{group}_{gene}.fasta",
-                        opt=opt,
-                        path=path,
-                        thread=opt.thread,
-                    )
+    # In case of IQTREE ModelFinder, for optimization, run it with multithreaded
+    elif opt.method.modeltest.lower() == "iqtree":
+        if not (opt.method.tree == "iqtree"):
+            if opt.verbose < 3:
+                modelfinder_opt = []
+                for group in group_dict:
+                    for gene in group_dict[group]:
+                        modelfinder_opt.append(
+                            (
+                                f"{path.out_alignment}/{opt.runname}_trimmed_{group}_{gene}.fasta",
+                                opt,
+                                path,
+                                1,
+                            )
+                        )
+
+                p = mp.Pool(opt.thread)
+                _ = p.starmap(ext.ModelFinder, modelfinder_opt)
+                p.close()
+                p.join()
+
+            else:  # non-multithreading mode for debugging
+                for group in group_dict:
+                    for gene in group_dict[group]:
+                        # Run IQTREE ModelFinder
+                        ext.ModelFinder(
+                            fasta=f"{path.out_alignment}/{opt.runname}_trimmed_{group}_{gene}.fasta",
+                            opt=opt,
+                            path=path,
+                            thread=opt.thread,
+                        )
 
-                    # Move modeltest result to appropriate location
+            # Move modeltest result to appropriate location
+            for group in group_dict:
+                for gene in group_dict[group]:
                     try:
                         shutil.move(
                             f"{path.out_alignment}/{opt.runname}_trimmed_{group}_{gene}.fasta.iqtree",
@@ -424,13 +451,15 @@ def modeltest(V, path, opt) -> dict:
                         path=path,
                         opt=opt,
                     )
-                else:
-                    logging.info(
-                        "IQTREE will perform ModelFinder internally in tree construction step, skipping in modeltest step"
-                    )
-                    model_dict[group][gene] = "skip"
+        else:
+            logging.info(
+                "IQTREE will perform ModelFinder internally in tree construction step, skipping in modeltest step"
+            )
+            model_dict[group][gene] = "skip"
 
-            else:  # including opt.model_method.lower() == "none":
+    else:  # including opt.model_method.lower() == "none":
+        for group in group_dict:
+            for gene in group_dict[group]:
                 if opt.method.tree == "raxml":
                     logging.info("Skipping modeltest. Using GTRGAMMA as default")
                     model_dict[group][gene] = "-m GTRGAMMA"

{funvip-0.3.24.5 → funvip-0.3.24.7}/funvip/src/tree.py

@@ -172,8 +172,7 @@ def pipe_tree(V, path, opt, model_dict):
             p.close()
             p.join()
 
-        else:
-            # non-multithreading mode for debugging
+        else:  # non-multithreading mode for debugging
             fasttree_result = []
             for option in fasttree_opt:
                 fasttree_result.append(ext.FastTree(*option))

{funvip-0.3.24.5 → funvip-0.3.24.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "FunVIP"
-version = "0.3.24.5"
+version = "0.3.24.7"
 description = "Fungal Validation & Identification Pipeline"
 authors = [{name = "Changwan Seo", email = "wan101010@snu.ac.kr"}]
 urls = { "Homepage" = "https://github.com/Changwanseo/FunVIP" }