remove_file fix and type fix

pcgr/arg_checker.py

@@ -75,18 +75,18 @@ def check_args(arg_dict):
         err_msg = f"Minimum sequencing depth tumor - TMB calculation ('tmb_dp_min' = {arg_dict['tmb_dp_min']}) must be >= 0"
         error_message(err_msg, logger)
 
-    if int(arg_dict['tmb_dp_min']) < int(arg_dict['tumor_dp_min']):
-        err_msg = f"Minimum sequencing depth (tumor) for TMB calculation ('tmb_dp_min' = {arg_dict['tmb_dp_min']}) must be ",
-        err_msg += f"greater or equal to minimum sequencing depth tumor {arg_dict['tumor_dp_min']} (i.e. filter for variant inclusion in report)"
+    if int(arg_dict['tmb_dp_min']) > 0 and (int(arg_dict['tmb_dp_min']) < int(arg_dict['tumor_dp_min'])):
+        err_msg = f"Minimum sequencing depth (tumor) for TMB calculation ('tmb_dp_min' = {str(arg_dict['tmb_dp_min'])}) must be ",
+        err_msg += f"greater or equal to minimum sequencing depth tumor {str(arg_dict['tumor_dp_min'])} (i.e. filter for variant inclusion in report)"
         error_message(err_msg, logger)
 
     if float(arg_dict['tmb_af_min']) < 0 or float(arg_dict['tmb_af_min']) > 1:
         err_msg = f"Minimum AF (tumor) for TMB calculation ('tmb_af_min' = {arg_dict['tmb_af_min']}) must be within [0, 1]"
         error_message(err_msg, logger)
 
-    if float(arg_dict['tmb_af_min']) < float(arg_dict['tumor_af_min']):
-        err_msg = f"Minimum AF (tumor) for TMB calculation ('tmb_af_min' = {arg_dict['tmb_af_min']}) must be ",
-        err_msg += f"greater or equal to minimum AF tumor {arg_dict['tumor_dp_min']} (i.e. filter for variant inclusion in report)"
+    if float(arg_dict['tmb_af_min']) > 0 and (float(arg_dict['tmb_af_min']) < float(arg_dict['tumor_af_min'])):
+        err_msg = f"Minimum AF (tumor) for TMB calculation ('tmb_af_min' = {str(arg_dict['tmb_af_min'])}) must be ",
+        err_msg += f"greater or equal to minimum AF tumor {str(arg_dict['tumor_dp_min'])} (i.e. filter for variant inclusion in report)"
         error_message(err_msg, logger)
 
     # Check that coding target size region of sequencing assay is set correctly

pcgr/cpsr.py

@@ -15,7 +15,7 @@
 from glob import glob
 from argparse import RawTextHelpFormatter
 from pcgr import pcgr_vars, arg_checker, utils, vep, config, variant
-from pcgr.utils import check_subprocess, getlogger, error_message, warn_message
+from pcgr.utils import check_subprocess, getlogger, error_message, warn_message, remove_file
 from pcgr.config import populate_config_data
 
 def get_args():
@@ -267,7 +267,7 @@ def run_cpsr(conf_options, cpsr_paths):
         # do not delete if debugging
         if not debug:
             for fn in delete_files:
-                utils.remove(fn)
+                remove_file(fn)
         logger.info('Finished cpsr-summarise main command')
 
         # CPSR|vcf2tsvpy - convert VCF to TSV with https://github.com/sigven/vcf2tsvpy
@@ -291,7 +291,7 @@ def run_cpsr(conf_options, cpsr_paths):
 
         variant_set.to_csv(output_pass_tsv_gz, sep="\t", compression="gzip", index=False)
         if not debug:
-            utils.remove(output_pass_vcf2tsv_gz)
+            remove_file(output_pass_vcf2tsv_gz)
 
         logger.info('Finished cpsr-summarise')
 

pcgr/utils.py

@@ -169,7 +169,7 @@ def sort_bed(unsorted_bed_fname: str, sorted_bed_fname: str, debug = False, logg
         check_subprocess(logger, cmd_sort_custom_bed1, debug)
         check_subprocess(logger, cmd_sort_custom_bed2, debug)
         if not debug:
-            remove(str(unsorted_bed_fname))
+            remove_file(str(unsorted_bed_fname))
     else:
         err_msg = 'File ' + str(unsorted_bed_fname) + ' does not exist or is empty'
         error_message(err_msg, logger)

pcgr/variant.py

@@ -286,10 +286,10 @@ def clean_annotations(variant_set: pd.DataFrame, yaml_data: dict, germline: bool
         for tag in ['AN','AC','NHOMALT']:
             vcf_info_tag = 'gnomADe_non_cancer_' + str(pop) + '_' + str(tag)
             if vcf_info_tag in variant_set.columns:
-                variant_set[vcf_info_tag] = variant_set[vcf_info_tag].astype(str)
+                #variant_set[vcf_info_tag] = variant_set[vcf_info_tag].astype(str)
 
                 variant_set.loc[variant_set[vcf_info_tag].notna(), vcf_info_tag] = \
-                    variant_set.loc[variant_set[vcf_info_tag].notna(), vcf_info_tag].astype(float).astype(int)
+                    variant_set.loc[variant_set[vcf_info_tag].notna(), vcf_info_tag].astype(str).astype(float).astype(int)
 
     for elem in ['NUM_SUBMITTERS','ALLELE_ID','ENTREZGENE','REVIEW_STATUS_STARS']:
         vcf_info_tag = 'CLINVAR_' + str(elem)

scripts/cpsr_validate_input.py

@@ -12,7 +12,7 @@
 
 from cyvcf2 import VCF
 from pcgr import utils, annoutils, vcf, pcgr_vars
-from pcgr.utils import error_message, check_subprocess, random_id_generator, sort_bed, check_file_exists
+from pcgr.utils import error_message, check_subprocess, random_id_generator, sort_bed, check_file_exists, remove_file
 
 
 def __main__():
@@ -262,10 +262,10 @@ def simplify_vcf(input_vcf, validated_vcf, vcf, custom_bed, pcgr_directory, geno
                    bcftools_simplify_log, 
                    vt_decompose_log]:
             #print(f"Deleting {fn}")
-            utils.remove(fn)
+            remove_file(fn)
 
-        utils.remove(temp_files["vcf_2"] + str('.tbi'))
-        utils.remove(temp_files["vcf_3"] + str('.tbi'))
+        remove_file(temp_files["vcf_2"] + str('.tbi'))
+        remove_file(temp_files["vcf_3"] + str('.tbi'))
 
     if check_file_exists(f'{validated_vcf}.gz'):
         vcf = VCF(validated_vcf + '.gz')