Feature/leeds crc new (#166)

* Updates to run two leeds hanlder pre and post 2025

* Leeds colorectal new handler FS

* reverting removal of TACSTD1 as Birmingham uses it

* Targ rules coded

* rubocop fix for safenavigation

* genes_panel method rubocopped

* More rubocop fixes

* tests added

* fixed cdna regex

* populating variantgenotype in table

* more refinements

* rubocop fixes

* more rubocop fixes

* rubocop fixes

* Pr Review comment done on versioning the handlers

* Remove older test files

* Alignment fixed

* Fix broken tests

Author: shilpigoeldev

lib/import/brca/providers/leeds/leeds_handler_new_format.rb

@@ -384,7 +384,6 @@ def classify_variant_pathogenicity(variantpathclass)
           end
 
           def classify_protein_impact
-            require 'pry'
             case @value1
             when /C1/
               1

lib/import/brca/scripts/bash/Import_all_brca_interactive.sh

@@ -54,7 +54,7 @@ MBIS=$1
 PROV='RR8'
 IFS=$'\n'
 for x in $(find $DIRPATH/$FILEPATH  -path "*/$PROV/*" -type f \
-\( -name "*BRCA*.pseudo" -o -type f -name "*Other*.pseudo" \) \
+\( -name "*BRCA*.pseudo" -o -type f -iname "*Other*.pseudo" \) \
 \( -path "*/202[5-9]/*" -o -path "*/203[0-9]/*" \) \
 ! -name "bede6d1385c0ae9db4fe61fe9b07d58f86e2dc60_24.11.2021 to 31.03.2025_BRCA_DATA__2021_11_24__to__2025_03_31_b.xlsx.pseudo")
 do

lib/import/colorectal/core/colorectal_handler_mapping.rb

@@ -4,7 +4,8 @@ module Core
       # Provides the handler appropriate for the dataformat from each center
       class ColorectalHandlerMapping
         HANDLER_MAPPING = {
-          'RR8' => Import::Colorectal::Providers::Leeds::LeedsHandlerColorectal,
+          'RR8_V1_PRE2025' => Import::Colorectal::Providers::Leeds::LeedsHandlerColorectalV1,
+          'RR8_V2_POST2025' => Import::Colorectal::Providers::Leeds::LeedsHandlerColorectalV2,
           'RNZ' => Import::Colorectal::Providers::Salisbury::SalisburyHandlerColorectal,
           'RTD' => Import::Colorectal::Providers::Newcastle::NewcastleHandlerColorectal,
           'RX1' => Import::Colorectal::Providers::Nottingham::NottinghamHandlerColorectal,

lib/import/colorectal/core/genocolorectal.rb

@@ -16,11 +16,16 @@ class Genocolorectal < Import::Germline::Genotype
         #--------------------- Schema code mapping tables --------------------------
 
         COLORECTAL_MAP = { 'APC' => 358,
+                           'BAP1' => 517,
                            'BMPR1A' => 577,
                            'EPCAM' => 1432,
-                           'TACSTD1' => 1432,
+                           'TACSTD1' => 1432, #Old symbol for EPCAM
+                           'FH' => 1590,
+                           'FLCN' => 1603,
+                           'MET' => 50,
                            'MLH1' => 2744,
                            'MSH2' => 2804,
+                           'MSH3' => 2805,
                            'MSH6' => 2808,
                            'MUTYH' => 2850,
                            'PMS2' => 3394,
@@ -42,14 +47,21 @@ class Genocolorectal < Import::Germline::Genotype
                            'RAD51D' => 3616,
                            'VHL' => 83,
                            'ATM' => 451,
-                           'SCG5' => 5092 }.freeze
+                           'SCG5' => 5092,
+                           'SDHB' => 68
+                         }.freeze
 
         COLORECTAL_REGEX = /(?<apc>APC)|
+                            (?<bap1>BAP1)|
                             (?<bmpr>BMPR1A)|
                             (?<epcam>EPCAM)|
-                            (?<tacstd1>TACSTD1)|
+                            (?<tacstd1>TACSTD1)| #Old symbol for EPCAM
+                            (?<fh>FH)|
+                            (?<flcn>FLCN)|
+                            (?<met>MET)|
                             (?<mlh1>MLH1)|
                             (?<msh2>MSH2)|
+                            (?<msh3>MSH3)|
                             (?<msh6>MSH6)|
                             (?<mutyh>MUTYH)|
                             (?<pms2>PMS2)|
@@ -71,15 +83,16 @@ class Genocolorectal < Import::Germline::Genotype
                             (?<rad51d>RAD51D)|
                             (?<vhl>VHL) |
                             (?<atm>ATM) |
-                            (?<scg5>SCG5)/ix # Added by Francesco
+                            (?<scg5>SCG5)|
+                            (?<sdhb>SDHB)/ix # Added by Francesco
 
         # ------------------------ Interogators ------------------------------
 
         def add_gene_colorectal(colorectal_input)
           case colorectal_input
           when Integer
-            if [1432, 358, 577, 2744, 2804, 2808, 2850, 3394, 7, 8, 79, 3186, 5019,
-                3408, 5000, 62, 72, 76, 1882, 3108, 794, 83, 5019, 451].include? colorectal_input
+            if [1432, 358, 517, 577, 2744, 2804, 2805, 2808, 2850, 3394, 7, 8, 79, 3186, 5019, 1603, 50, 68,
+                3408, 5000, 62, 72, 76, 1882, 3108, 794, 83, 5019, 451, 1590, 5092].include? colorectal_input
 
               @attribute_map['gene'] = colorectal_input
               @logger.debug "SUCCESSFUL gene parse for #{colorectal_input}"

…viders/leeds/leeds_handler_colorectal.rb …ers/leeds/leeds_handler_colorectal_v1.rblib/import/colorectal/providers/leeds/leeds_handler_colorectal.rb renamed to lib/import/colorectal/providers/leeds/leeds_handler_colorectal_v1.rb lib/import/colorectal/providers/leeds/leeds_handler_colorectal.rb renamed to lib/import/colorectal/providers/leeds/leeds_handler_colorectal_v1.rb

@@ -3,8 +3,8 @@ module Colorectal
     module Providers
       module Leeds
         # rubocop:disable Metrics/ClassLength
-        # Leeds importer for colorectal
-        class LeedsHandlerColorectal < Import::Germline::ProviderHandler
+        # Leeds importer for colorectal (pre-2025 format)
+        class LeedsHandlerColorectalV1 < Import::Germline::ProviderHandler
           include Import::Helpers::Colorectal::Providers::Rr8::Constants
 
           def initialize(batch)
@@ -53,6 +53,9 @@ def populate_and_persist_genotype(record)
             add_varclass
             add_organisationcode_testresult(genocolorectal)
             res = process_variants_from_record(genocolorectal, record)
+            # correcting ebatch provider and registry to RR8 (from RR8_V1_PRE2025) to allow data to persist in the database
+            @batch.provider = 'RR8'
+            @batch.registryid = 'RR8'
             res.map { |cur_genotype| @persister.integrate_and_store(cur_genotype) }
           end