From 439731a544dd6f62d6602ee9fbae8813a7e6372d Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 20 Aug 2024 11:46:14 +0100
Subject: [PATCH 01/14] Capture uploaded allele correctly for VCF input

---
 modules/Bio/EnsEMBL/VEP/OutputFactory.pm | 3 ++-
 modules/Bio/EnsEMBL/VEP/Parser.pm        | 2 +-
 modules/Bio/EnsEMBL/VEP/Parser/VCF.pm    | 9 +++++++++
 3 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/OutputFactory.pm b/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
index 70cd5456e..a5d5da5ee 100755
--- a/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
+++ b/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
@@ -1296,7 +1296,8 @@ sub VariationFeatureOverlapAllele_to_output_hash {
 
   # reference allele
   $hash->{REF_ALLELE} = $vf->ref_allele_string if $self->{show_ref_allele};
- 
+
+  # Capture uploaded allele
   $hash->{UPLOADED_ALLELE} = ($vf->{original_allele_string} || $vf->{allele_string} || $vf->{class_SO_term} || "" ) if $self->param('uploaded_allele');
 
   # picked?
diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 5270770f4..2a5c77f78 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -976,7 +976,7 @@ sub minimise_alleles {
         $new_vf->{end}                    = $end;
         $new_vf->{seq_region_start}       = $start;
         $new_vf->{seq_region_end}         = $end;
-        $new_vf->{original_allele_string} = $vf->{allele_string};
+        $new_vf->{original_allele_string} = $vf->{original_allele_string} || $vf->{allele_string};
         $new_vf->{original_start}         = $vf->{start};
         $new_vf->{original_end}           = $vf->{end};
         $new_vf->{minimised}              = 1;
diff --git a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
index ee9a46fbe..8f1bd37dc 100644
--- a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
@@ -287,6 +287,9 @@ sub create_VariationFeatures {
   # record original alleles
   # if they get changed, we need to map from old to new in create_individual_VariationFeatures
   my @original_alleles = ($ref, @$alts);
+  use Data::Dumper;
+  print("Original alleles"."\n");
+  print(Dumper(@original_alleles));
 
   # adjust end coord
   # $end += (length($ref) - 1);
@@ -336,10 +339,14 @@ sub create_VariationFeatures {
   }
 
   # create VF object
+  use Data::Dumper;
+  # print("VF:".$ref."\n");
+  # print("VF:".Dumper(@$alts)."\n");
   my $vf = Bio::EnsEMBL::Variation::VariationFeature->new_fast({
     start          => $start,
     end            => $end,
     allele_string  => $non_variant ? $ref : $ref.'/'.join('/', @$alts),
+    original_allele_string => $non_variant ? $original_alleles[0] : join("/",@original_alleles),
     strand         => 1,
     map_weight     => 1,
     adaptor        => $self->get_adaptor('variation', 'VariationFeature'),
@@ -348,6 +355,8 @@ sub create_VariationFeatures {
     _line          => $record,
   });
 
+  print($vf->{original_allele_string});
+
   # flag as non-variant
   $vf->{non_variant} = 1 if $non_variant;
 

From 11adcb8c0bba93c8abd4623a0b0687aae0c98f00 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 20 Aug 2024 11:48:12 +0100
Subject: [PATCH 02/14] Cleanup

---
 modules/Bio/EnsEMBL/VEP/Parser/VCF.pm | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
index 8f1bd37dc..5706fdbe1 100644
--- a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
@@ -287,9 +287,6 @@ sub create_VariationFeatures {
   # record original alleles
   # if they get changed, we need to map from old to new in create_individual_VariationFeatures
   my @original_alleles = ($ref, @$alts);
-  use Data::Dumper;
-  print("Original alleles"."\n");
-  print(Dumper(@original_alleles));
 
   # adjust end coord
   # $end += (length($ref) - 1);
@@ -339,9 +336,6 @@ sub create_VariationFeatures {
   }
 
   # create VF object
-  use Data::Dumper;
-  # print("VF:".$ref."\n");
-  # print("VF:".Dumper(@$alts)."\n");
   my $vf = Bio::EnsEMBL::Variation::VariationFeature->new_fast({
     start          => $start,
     end            => $end,
@@ -355,7 +349,6 @@ sub create_VariationFeatures {
     _line          => $record,
   });
 
-  print($vf->{original_allele_string});
 
   # flag as non-variant
   $vf->{non_variant} = 1 if $non_variant;

From 3538ac4fa86d69dc780aa09c9ef1e25a90828cb6 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 20 Aug 2024 11:49:06 +0100
Subject: [PATCH 03/14] Cleanup

---
 modules/Bio/EnsEMBL/VEP/Parser/VCF.pm | 1 -
 1 file changed, 1 deletion(-)

diff --git a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
index 5706fdbe1..3c8d9277c 100644
--- a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
@@ -349,7 +349,6 @@ sub create_VariationFeatures {
     _line          => $record,
   });
 
-
   # flag as non-variant
   $vf->{non_variant} = 1 if $non_variant;
 

From 5f1d345115fa629c819ac8945cc60eec180b8e02 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 20 Aug 2024 16:52:02 +0100
Subject: [PATCH 04/14] Supporting uploaded allele for multiple alternate
 alleles

---
 modules/Bio/EnsEMBL/VEP/InputBuffer.pm | 2 ++
 modules/Bio/EnsEMBL/VEP/Parser.pm      | 7 ++++++-
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
index 617cc35b0..46458bcd8 100644
--- a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
+++ b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -209,6 +209,8 @@ sub next {
   if(my $parser = $self->parser) {
     while(@$buffer < $buffer_size && (my $vf = $parser->next)) {
 
+      ## Set minimal to 1 if indel
+      $self->{minimal} = 1 if (defined($vf->{minimised}) && $vf->{minimised});
 
       # exit the program if the maximum number of variants not ordered in the input file is reached
       if (!$self->param('no_check_variants_order') &&
diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 2a5c77f78..870c12619 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -950,10 +950,15 @@ sub minimise_alleles {
 
     # skip VFs with more than one alt
     # they get taken care of later by split_variants/rejoin_variants
-    if(!$vf->{allele_string} || $vf->{allele_string} =~ /.+\/.+\/.+/ || $vf->{allele_string} !~ /.+\/.+/) {
+    if(!$vf->{allele_string} || $vf->{allele_string} !~ /.+\/.+/) {
       push @return, $vf;
     }
 
+    elsif($vf->{allele_string} =~ /.+\/.+\/.+/)
+    {
+      $vf->{minimised} = 1;
+    }
+
     else {
       my @alleles = split('/', $vf->{allele_string});
       my $ref = shift @alleles;

From 6794d2451f2643be6c4b9de35b45dd6791144b7f Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Wed, 21 Aug 2024 22:22:52 +0100
Subject: [PATCH 05/14] Return vf after enabling minimised

---
 modules/Bio/EnsEMBL/VEP/Parser.pm | 1 +
 1 file changed, 1 insertion(+)

diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 870c12619..142432ac3 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -957,6 +957,7 @@ sub minimise_alleles {
     elsif($vf->{allele_string} =~ /.+\/.+\/.+/)
     {
       $vf->{minimised} = 1;
+      push @return, $vf;
     }
 
     else {

From c039ac7da92f33498fc6d68583e692cb2bb88a7d Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Thu, 22 Aug 2024 12:54:35 +0100
Subject: [PATCH 06/14] Clearer variable name and comments

---
 modules/Bio/EnsEMBL/VEP/Parser.pm | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 142432ac3..8f3c1b038 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -851,13 +851,13 @@ sub post_process_vfs {
     $vf->seq_region_start($vf->{start});
     $vf->seq_region_end($vf->{end});
 
-    # Checks if the allele string is insertion or/and deletion
+    # Checks if the allele string is non-minimised insertion or/and deletion
     my $is_sv = ref($vf) eq 'Bio::EnsEMBL::Variation::StructuralVariationFeature';
     if(!$is_sv && defined($vf->{allele_string}) && $vf->{allele_string} =~ /\//){
-      my $is_indel = 0;
+      my $is_non_minimised_indel = 0;
       my ($ref_allele_string,$alt_allele_string) = split(/\//, $vf->{allele_string});
-      $is_indel = 1 unless length($ref_allele_string) == length($alt_allele_string) or $vf->{allele_string} =~ /-/;
-      $vf = ${$self->minimise_alleles([$vf])}[0] if $is_indel;
+      $is_non_minimised_indel = 1 unless length($ref_allele_string) == length($alt_allele_string) or $vf->{allele_string} =~ /-/;
+      $vf = ${$self->minimise_alleles([$vf])}[0] if $is_non_minimised_indel;
     }
   }
   return $vfs;

From b7b2013a8550cfc435157cb67b7d681334e744aa Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Fri, 30 Aug 2024 17:44:02 +0100
Subject: [PATCH 07/14] Renaming conflicting variable names

---
 modules/Bio/EnsEMBL/VEP/InputBuffer.pm |  3 +++
 modules/Bio/EnsEMBL/VEP/Parser.pm      | 10 +++++++---
 modules/Bio/EnsEMBL/VEP/Parser/VCF.pm  |  2 +-
 3 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
index 46458bcd8..1d202ffc8 100644
--- a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
+++ b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -191,6 +191,9 @@ sub next {
   while(@$pre_buffer && @$buffer < $buffer_size) {
     my $vf = $pre_buffer->[0];
 
+    ## Set minimal to 1 if indel
+    $self->{minimal} = 1 if (defined($vf->{minimised}) && $vf->{minimised});
+
     # new chromosome
     if($prev_chr && $vf->{chr} ne $prev_chr) {
       $self->split_variants() if $self->{minimal};
diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 8f3c1b038..f5156ee29 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -855,8 +855,11 @@ sub post_process_vfs {
     my $is_sv = ref($vf) eq 'Bio::EnsEMBL::Variation::StructuralVariationFeature';
     if(!$is_sv && defined($vf->{allele_string}) && $vf->{allele_string} =~ /\//){
       my $is_non_minimised_indel = 0;
-      my ($ref_allele_string,$alt_allele_string) = split(/\//, $vf->{allele_string});
-      $is_non_minimised_indel = 1 unless length($ref_allele_string) == length($alt_allele_string) or $vf->{allele_string} =~ /-/;
+      # For VCF input, the allele_string is trimmed to remove anchor base, so we capture original allele
+      
+      my $original_allele_string = ($vf->{nontrimmed_allele_string} || $vf->{allele_string});
+      my ($ref_allele_string,$alt_allele_string) = split(/\//, $original_allele_string);
+      $is_non_minimised_indel = 1 unless length($ref_allele_string) == length($alt_allele_string) or $original_allele_string =~ /-/;
       $vf = ${$self->minimise_alleles([$vf])}[0] if $is_non_minimised_indel;
     }
   }
@@ -956,6 +959,7 @@ sub minimise_alleles {
 
     elsif($vf->{allele_string} =~ /.+\/.+\/.+/)
     {
+      # Updating a flag to minimise multi-allelic variants in split_variants/rejoin_variants
       $vf->{minimised} = 1;
       push @return, $vf;
     }
@@ -982,7 +986,7 @@ sub minimise_alleles {
         $new_vf->{end}                    = $end;
         $new_vf->{seq_region_start}       = $start;
         $new_vf->{seq_region_end}         = $end;
-        $new_vf->{original_allele_string} = $vf->{original_allele_string} || $vf->{allele_string};
+        $new_vf->{original_allele_string} = $vf->{nontrimmed_allele_string} || $vf->{allele_string};
         $new_vf->{original_start}         = $vf->{start};
         $new_vf->{original_end}           = $vf->{end};
         $new_vf->{minimised}              = 1;
diff --git a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
index 3c8d9277c..8e58bcf71 100644
--- a/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser/VCF.pm
@@ -340,7 +340,7 @@ sub create_VariationFeatures {
     start          => $start,
     end            => $end,
     allele_string  => $non_variant ? $ref : $ref.'/'.join('/', @$alts),
-    original_allele_string => $non_variant ? $original_alleles[0] : join("/",@original_alleles),
+    nontrimmed_allele_string => $non_variant ? $original_alleles[0] : join("/",@original_alleles),
     strand         => 1,
     map_weight     => 1,
     adaptor        => $self->get_adaptor('variation', 'VariationFeature'),

From 8ad71eb1c65006f33931767549fd026658ba1eac Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 5 Nov 2024 13:59:17 +0000
Subject: [PATCH 08/14] Fix buffer logic for multi-allelic variants

---
 modules/Bio/EnsEMBL/VEP/InputBuffer.pm |  8 +-------
 modules/Bio/EnsEMBL/VEP/Parser.pm      | 13 +++++++++++--
 2 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
index 1d202ffc8..27db09296 100644
--- a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
+++ b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -191,9 +191,6 @@ sub next {
   while(@$pre_buffer && @$buffer < $buffer_size) {
     my $vf = $pre_buffer->[0];
 
-    ## Set minimal to 1 if indel
-    $self->{minimal} = 1 if (defined($vf->{minimised}) && $vf->{minimised});
-
     # new chromosome
     if($prev_chr && $vf->{chr} ne $prev_chr) {
       $self->split_variants() if $self->{minimal};
@@ -212,8 +209,6 @@ sub next {
   if(my $parser = $self->parser) {
     while(@$buffer < $buffer_size && (my $vf = $parser->next)) {
 
-      ## Set minimal to 1 if indel
-      $self->{minimal} = 1 if (defined($vf->{minimised}) && $vf->{minimised});
 
       # exit the program if the maximum number of variants not ordered in the input file is reached
       if (!$self->param('no_check_variants_order') &&
@@ -262,8 +257,7 @@ sub next {
     die($error_msg);
   }
 
-  $self->split_variants() if $self->{minimal};
-
+  $self->split_variants() if $self->{minimal}  ;
   return $buffer;
 }
 
diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index f5156ee29..31c159e3b 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -858,8 +858,16 @@ sub post_process_vfs {
       # For VCF input, the allele_string is trimmed to remove anchor base, so we capture original allele
       
       my $original_allele_string = ($vf->{nontrimmed_allele_string} || $vf->{allele_string});
-      my ($ref_allele_string,$alt_allele_string) = split(/\//, $original_allele_string);
-      $is_non_minimised_indel = 1 unless length($ref_allele_string) == length($alt_allele_string) or $original_allele_string =~ /-/;
+      my @alleles = split(/\//, $original_allele_string);
+      my $ref_allele_string = shift @alleles;
+      my $alt_allele_count;
+
+      foreach my $alt(@alleles) {
+        if (length($ref_allele_string) != length($alt) or $original_allele_string =~ /^-/){
+          $is_non_minimised_indel = 1;
+          last;
+        }
+      }
       $vf = ${$self->minimise_alleles([$vf])}[0] if $is_non_minimised_indel;
     }
   }
@@ -961,6 +969,7 @@ sub minimise_alleles {
     {
       # Updating a flag to minimise multi-allelic variants in split_variants/rejoin_variants
       $vf->{minimised} = 1;
+      $vf->{original_allele_string} = $vf->{nontrimmed_allele_string} || $vf->{allele_string};
       push @return, $vf;
     }
 

From d3d655cd55e9fb7157393d6878d0822b01e0b802 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Wed, 4 Dec 2024 13:08:49 +0000
Subject: [PATCH 09/14] Modify unittests

---
 t/Parser_VCF.t | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/t/Parser_VCF.t b/t/Parser_VCF.t
index 6fa38488c..cc37d2671 100755
--- a/t/Parser_VCF.t
+++ b/t/Parser_VCF.t
@@ -72,6 +72,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'rs142513484',
   'map_weight' => 1,
   'allele_string' => 'C/T',
+  'nontrimmed_allele_string' => 'C/T',
   'end' => 25585733,
   'start' => 25585733,
   'seq_region_end' => 25585733,
@@ -115,6 +116,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'C/-',
+  'minimised' => 1,
   'end' => 25587760,
   'start' => 25587760,
   'seq_region_end' => 25587760,
@@ -133,6 +135,7 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => '-/C',
   'end' => 25587759,
   'start' => 25587760,
@@ -152,6 +155,7 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => 'C/-',
   'end' => 25587760,
   'start' => 25587760,
@@ -171,6 +175,7 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => '-/C',
   'end' => 25587759,
   'start' => 25587760,
@@ -191,6 +196,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'A/C/G',
+  'nontrimmed_allele_string' => 'A/C/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -210,6 +216,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'A/C/GG',
+  'original_allele_string' => 'A/C/GG',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -229,6 +236,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => '-/C/T',
+  'original_allele_string' => 'A/C/GG',
   'end' => 25587759,
   'start' => 25587760,
   'seq_region_end' => 25587759,
@@ -248,6 +256,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'G/C',
+  'nontrimmed_allele_string' => 'G/C',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -267,6 +276,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'G',
+  'nontrimmed_allele_string' => 'G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -294,6 +304,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'G/C/*',
+  'nontrimmed_allele_string' => 'G/C/*',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -312,6 +323,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'G/C/<DEL:*>',
+  'original_allele_string' => 'G/C/<DEL:*>',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -331,6 +343,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'C/-/*',
+  'original_allele_string' => 'GC/G/*',
   'end' => 25587760,
   'start' => 25587760,
   'seq_region_end' => 25587760,
@@ -350,6 +363,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => '-/C/*',
+  'original_allele_string' => 'G/GC/*',
   'end' => 25587759,
   'start' => 25587760,
   'seq_region_end' => 25587759,
@@ -393,6 +407,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'C/T/CAA',
+  'original_allele_string' => 'C/T/CAA',
   'end' => 25587758,
   'start' => 25587758,
   'seq_region_end' => 25587758,
@@ -557,6 +572,7 @@ is_deeply($snv, bless( {
   'chr' => '21',
   'strand' => 1,
   'variation_name' => 'test',
+  'nontrimmed_allele_string' => 'A/C',
   'map_weight' => 1,
   'allele_string' => 'A/C',
   'end' => 25587759,
@@ -923,6 +939,7 @@ is_deeply($tandem_RUC, bless( {
     'strand' => '1',
     'variation_name' => 'tr0',
     'allele_string' => 'AGTAAATAGA/' . 'CAT' x 2 . 'GT' x 5 . '/' . 'CA' x 4,
+    'original_allele_string' => 'AGTAAATAGA/CATCATGTGTGTGTGT/CACACACA',
     'start' => 25587760,
     'end' => 25587769,
     'seq_region_start' => 25587760,
@@ -941,6 +958,7 @@ is_deeply($tandem, bless( {
     'strand' => '1',
     'variation_name' => 'tr0',
     'allele_string' => 'AGTAAATAGA/' . 'CAT' x 2 . 'N' x 5 . '/' . 'CA' x 4,
+    'original_allele_string' => 'AGTAAATAGA/CATCATNNNNN/CACACACA',
     'start' => 25587760,
     'end' => 25587769,
     'seq_region_start' => 25587760,
@@ -964,6 +982,7 @@ is_deeply($tandem, bless( {
     'strand' => '1',
     'variation_name' => 'tr0',
     'allele_string' => 'A/' . 'CAT' x 2 . 'N' x 5 . '/' . 'CA' x 4,
+    'original_allele_string' => 'A/CATCATNNNNN/CACACACA',
     'start' => 25587760,
     'end' => 25587760,
     'seq_region_start' => 25587760,
@@ -1014,6 +1033,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'A/G',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25586000,
   'start' => 25586000,
   'seq_region_end' => 25586000,
@@ -1056,6 +1076,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'indtest',
   'map_weight' => 1,
   'allele_string' => 'A/G',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -1073,6 +1094,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'indtest',
   'map_weight' => 1,
   'allele_string' => 'A/G',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -1090,6 +1112,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'indtest',
   'map_weight' => 1,
   'allele_string' => 'A',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -1120,6 +1143,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'indtest',
   'map_weight' => 1,
   'allele_string' => 'A/A',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -1186,6 +1210,7 @@ is_deeply($vf, bless( {
   'variation_name' => 'indtest',
   'map_weight' => 1,
   'allele_string' => 'A/G',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,

From df56c6119056d6875ba9bbdbd62a13151c162eaf Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Wed, 4 Dec 2024 15:31:30 +0000
Subject: [PATCH 10/14] Changes in rejoin_vf; Fix unit tests

---
 modules/Bio/EnsEMBL/VEP/InputBuffer.pm   |  3 ++-
 modules/Bio/EnsEMBL/VEP/OutputFactory.pm |  3 ++-
 t/InputBuffer.t                          | 10 +++++++++
 t/OutputFactory.t                        |  4 +++-
 t/OutputFactory_JSON.t                   |  2 +-
 t/Parser_Region.t                        |  2 ++
 t/Parser_VCF.t                           | 26 +++++++++++++++++++-----
 7 files changed, 41 insertions(+), 9 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
index 27db09296..eb93f00ca 100644
--- a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
+++ b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -663,7 +663,8 @@ sub split_variants {
           $first->{original_allele_string} = $original_vf->{allele_string};
           $first->{original_start}         = $original_vf->{start};
           $first->{original_end}           = $original_vf->{end};
-          $first->{minimised}              = 1
+          $first->{minimised}              = 1;
+          $first->{first}              = 1;
         }
 
         push @tmp, $new_vf;
diff --git a/modules/Bio/EnsEMBL/VEP/OutputFactory.pm b/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
index a5d5da5ee..38332c179 100755
--- a/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
+++ b/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
@@ -2194,7 +2194,8 @@ sub rejoin_variants_in_InputBuffer {
   foreach my $vf(@{$buffer->buffer}) {
 
     # reset original one
-    if(defined($vf->{original_allele_string})) {
+    # check $vf->{first} to get first $vf in split_variants
+    if(defined($vf->{first})) {
 
       # do consequence stuff
       $self->get_all_output_hashes_by_VariationFeature($vf);
diff --git a/t/InputBuffer.t b/t/InputBuffer.t
index 8deb36850..47fba3220 100644
--- a/t/InputBuffer.t
+++ b/t/InputBuffer.t
@@ -72,6 +72,7 @@ is_deeply($vfs->[0], bless( {
   'variation_name' => 'rs142513484',
   'map_weight' => 1,
   'allele_string' => 'C/T',
+  'nontrimmed_allele_string' => 'C/T',
   'end' => 25585733,
   'start' => 25585733,
   'seq_region_end' => 25585733,
@@ -90,6 +91,7 @@ is_deeply($vfs->[0], bless( {
   'variation_name' => 'rs148490508',
   'map_weight' => 1,
   'allele_string' => 'A/G',
+  'nontrimmed_allele_string' => 'A/G',
   'end' => 25592911,
   'start' => 25592911,
   'seq_region_end' => 25592911,
@@ -418,7 +420,9 @@ is_deeply($ib->buffer, [
     'chr' => '1',
     'minimised' => 1,
     'original_allele_string' => 'CAGAAGAAAG/TAGAAGAAAG/C',
+    'nontrimmed_allele_string' => 'CAGAAGAAAG/TAGAAGAAAG/C',
     'original_end' => 10,
+    'first' => 1,
     'end' => 1,
     'original_start' => 1,
     'strand' => 1,
@@ -439,7 +443,10 @@ is_deeply($ib->buffer, [
     'variation_name' => '.',
     'alt_allele' => '-',
     'map_weight' => 1,
+    'minimised' => 1,
     'allele_string' => 'AGAAGAAAG/-',
+    'original_allele_string' => 'CAGAAGAAAG/TAGAAGAAAG/C',
+    'nontrimmed_allele_string' => 'CAGAAGAAAG/TAGAAGAAAG/C',
     'start' => 2,
     'seq_region_start' => 2,
     'seq_region_end' => 10,
@@ -470,7 +477,10 @@ is_deeply(
     'strand' => 1,
     'variation_name' => '.',
     'map_weight' => 1,
+    'minimised' => 1,
     'allele_string' => 'CAG/TAG/T',
+    'original_allele_string' => 'CAG/TAG/T',
+    'nontrimmed_allele_string' => 'CAG/TAG/T',
     'end' => 3,
     'start' => 1,
     'seq_region_start' => 1,
diff --git a/t/OutputFactory.t b/t/OutputFactory.t
index 665151aea..31e750ab5 100644
--- a/t/OutputFactory.t
+++ b/t/OutputFactory.t
@@ -1678,7 +1678,9 @@ $ib = get_annotated_buffer({
 
 is(scalar @{$ib->buffer}, 2, 'minimal - expanded count');
 is($ib->buffer->[0]->allele_string, 'C/T', 'minimal - expanded first allele string');
-
+# print("Before rejoin\n");
+# use Data::Dumper;
+# print(Dumper($ib->buffer));
 $of->rejoin_variants_in_InputBuffer($ib);
 
 is(scalar @{$ib->buffer}, 1, 'minimal - rejoined count');
diff --git a/t/OutputFactory_JSON.t b/t/OutputFactory_JSON.t
index 6ee7efd8a..86ff4871d 100755
--- a/t/OutputFactory_JSON.t
+++ b/t/OutputFactory_JSON.t
@@ -532,7 +532,7 @@ SKIP: {
       'cdna_start' => 2347,
       'transcript_id' => 'NM_000484.3',
       'gene_id' => '351',
-      'uploaded_allele' => '-/A',
+      'uploaded_allele' => 'G/GA',
       'cds_start' => 2147,
       'protein_start' => 716,
       'refseq_match' => [
diff --git a/t/Parser_Region.t b/t/Parser_Region.t
index 3efcdde05..4085ed712 100644
--- a/t/Parser_Region.t
+++ b/t/Parser_Region.t
@@ -127,7 +127,9 @@ is_deeply($vf, bless( {
   'strand' => '1',
   'variation_name' => '21:25587759-25587758:1/A',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => '-/A',
+  'original_allele_string' => '-/A',
   'end' => '25587758',
   'start' => '25587759',
   'seq_region_end' => '25587758',
diff --git a/t/Parser_VCF.t b/t/Parser_VCF.t
index cc37d2671..33586bfdd 100755
--- a/t/Parser_VCF.t
+++ b/t/Parser_VCF.t
@@ -215,7 +215,9 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => 'A/C/GG',
+  'nontrimmed_allele_string' => 'A/C/GG',
   'original_allele_string' => 'A/C/GG',
   'end' => 25587759,
   'start' => 25587759,
@@ -223,7 +225,7 @@ is_deeply($vf, bless( {
   'seq_region_start' => 25587759
 }, 'Bio::EnsEMBL::Variation::VariationFeature' ), 'mixed types - different first base');
 
-# mixed types - different first base
+# mixed types - same first base
 $vf = Bio::EnsEMBL::VEP::Parser::VCF->new({
   config => $cfg,
   file => $test_cfg->create_input_file([qw(21 25587759 test G GC,GT . . .)]),
@@ -235,8 +237,10 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => '-/C/T',
-  'original_allele_string' => 'A/C/GG',
+  'original_allele_string' => 'G/GC/GT',
+  'nontrimmed_allele_string' => 'G/GC/GT',
   'end' => 25587759,
   'start' => 25587760,
   'seq_region_end' => 25587759,
@@ -322,8 +326,10 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => 'G/C/<DEL:*>',
   'original_allele_string' => 'G/C/<DEL:*>',
+  'nontrimmed_allele_string' => 'G/C/<DEL:*>',
   'end' => 25587759,
   'start' => 25587759,
   'seq_region_end' => 25587759,
@@ -342,8 +348,10 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => 'C/-/*',
   'original_allele_string' => 'GC/G/*',
+  'nontrimmed_allele_string' => 'GC/G/*',
   'end' => 25587760,
   'start' => 25587760,
   'seq_region_end' => 25587760,
@@ -362,8 +370,10 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => '-/C/*',
   'original_allele_string' => 'G/GC/*',
+  'nontrimmed_allele_string' => 'G/GC/*',
   'end' => 25587759,
   'start' => 25587760,
   'seq_region_end' => 25587759,
@@ -382,6 +392,7 @@ is_deeply($vf, bless( {
   'chr' => '21',
   'minimised' => 1,
   'original_allele_string' => 'CAT/CCT',
+  'nontrimmed_allele_string' => 'CAT/CCT',
   'original_end' => 25587760,
   'end' => 25587759,
   'seq_region_end' => 25587759,
@@ -406,8 +417,10 @@ is_deeply($vf, bless( {
   'strand' => 1,
   'variation_name' => 'test',
   'map_weight' => 1,
+  'minimised' => 1,
   'allele_string' => 'C/T/CAA',
   'original_allele_string' => 'C/T/CAA',
+  'nontrimmed_allele_string' => 'C/T/CAA',
   'end' => 25587758,
   'start' => 25587758,
   'seq_region_end' => 25587758,
@@ -944,7 +957,8 @@ is_deeply($tandem_RUC, bless( {
     'end' => 25587769,
     'seq_region_start' => 25587760,
     'seq_region_end' => 25587769,
-    'map_weight' => 1
+    'map_weight' => 1,
+    'minimised' => 1
   },
   'Bio::EnsEMBL::Variation::VariationFeature' ) ,
   'VariationFeature - tandem repeat using RUC');
@@ -963,7 +977,8 @@ is_deeply($tandem, bless( {
     'end' => 25587769,
     'seq_region_start' => 25587760,
     'seq_region_end' => 25587769,
-    'map_weight' => 1
+    'map_weight' => 1,
+    'minimised' => 1
   },
   'Bio::EnsEMBL::Variation::VariationFeature' ) ,
   'VariationFeature - tandem repeat with missing sequence');
@@ -987,7 +1002,8 @@ is_deeply($tandem, bless( {
     'end' => 25587760,
     'seq_region_start' => 25587760,
     'seq_region_end' => 25587760,
-    'map_weight' => 1
+    'map_weight' => 1,
+    'minimised' => 1
   },
   'Bio::EnsEMBL::Variation::VariationFeature' ) ,
   'VariationFeature - tandem repeat with missing END and SVLEN');

From 56d965a6dd5bf23205c17ae10639a40434f9d322 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Thu, 5 Dec 2024 11:03:07 +0000
Subject: [PATCH 11/14] Fixed unit tests

---
 t/Parser_Region.t |  2 ++
 t/Parser_VCF.t    | 16 ++++++++++++++++
 2 files changed, 18 insertions(+)

diff --git a/t/Parser_Region.t b/t/Parser_Region.t
index 4085ed712..04763f892 100644
--- a/t/Parser_Region.t
+++ b/t/Parser_Region.t
@@ -132,6 +132,8 @@ is_deeply($vf, bless( {
   'original_allele_string' => '-/A',
   'end' => '25587758',
   'start' => '25587759',
+  'original_start' => '25587759',
+  'original_end' => '25587758',
   'seq_region_end' => '25587758',
   'seq_region_start' => '25587759',
   '_line' => ['21:25587759-25587758:1/A']
diff --git a/t/Parser_VCF.t b/t/Parser_VCF.t
index 33586bfdd..36004c0fe 100755
--- a/t/Parser_VCF.t
+++ b/t/Parser_VCF.t
@@ -116,9 +116,13 @@ is_deeply($vf, bless( {
   'variation_name' => 'test',
   'map_weight' => 1,
   'allele_string' => 'C/-',
+  'original_allele_string' => 'AC/A',
+  'nontrimmed_allele_string' => 'AC/A',
   'minimised' => 1,
   'end' => 25587760,
   'start' => 25587760,
+  'original_end' => 25587760,
+  'original_start' => 25587760,
   'seq_region_end' => 25587760,
   'seq_region_start' => 25587760
 }, 'Bio::EnsEMBL::Variation::VariationFeature' ), 'deletion');
@@ -137,8 +141,12 @@ is_deeply($vf, bless( {
   'map_weight' => 1,
   'minimised' => 1,
   'allele_string' => '-/C',
+  'original_allele_string' => 'A/AC',
+  'nontrimmed_allele_string' => 'A/AC',
   'end' => 25587759,
   'start' => 25587760,
+  'original_start' => 25587760,
+  'original_end' => 25587759,
   'seq_region_end' => 25587759,
   'seq_region_start' => 25587760
 }, 'Bio::EnsEMBL::Variation::VariationFeature' ), 'insertion');
@@ -157,8 +165,12 @@ is_deeply($vf, bless( {
   'map_weight' => 1,
   'minimised' => 1,
   'allele_string' => 'C/-',
+  'original_allele_string' => 'AC/A',
+  'nontrimmed_allele_string' => 'AC/A',
   'end' => 25587760,
   'start' => 25587760,
+  'original_end' => 25587760,
+  'original_start' => 25587760,
   'seq_region_end' => 25587760,
   'seq_region_start' => 25587760
 }, 'Bio::EnsEMBL::Variation::VariationFeature' ), 'deletion ignore SVTYPE');
@@ -177,8 +189,12 @@ is_deeply($vf, bless( {
   'map_weight' => 1,
   'minimised' => 1,
   'allele_string' => '-/C',
+  'original_allele_string' => 'A/AC',
+  'nontrimmed_allele_string' => 'A/AC',
   'end' => 25587759,
   'start' => 25587760,
+  'original_end' => 25587759,
+  'original_start' => 25587760,
   'seq_region_end' => 25587759,
   'seq_region_start' => 25587760
 }, 'Bio::EnsEMBL::Variation::VariationFeature' ), 'insertion ignore SVLEN');

From a8fbad34fd7d303e623fe75547ce37501fb736bb Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Tue, 17 Dec 2024 13:46:00 +0000
Subject: [PATCH 12/14] Modified non_minimised_indel logic; code cleanup

---
 modules/Bio/EnsEMBL/VEP/Parser.pm | 6 +++++-
 t/OutputFactory.t                 | 4 +---
 2 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/Parser.pm b/modules/Bio/EnsEMBL/VEP/Parser.pm
index 31c159e3b..af01711d9 100755
--- a/modules/Bio/EnsEMBL/VEP/Parser.pm
+++ b/modules/Bio/EnsEMBL/VEP/Parser.pm
@@ -863,7 +863,11 @@ sub post_process_vfs {
       my $alt_allele_count;
 
       foreach my $alt(@alleles) {
-        if (length($ref_allele_string) != length($alt) or $original_allele_string =~ /^-/){
+        if ($original_allele_string =~ /-/)
+        {
+          last;
+        }
+        elsif (length($ref_allele_string) != length($alt)) {
           $is_non_minimised_indel = 1;
           last;
         }
diff --git a/t/OutputFactory.t b/t/OutputFactory.t
index 31e750ab5..665151aea 100644
--- a/t/OutputFactory.t
+++ b/t/OutputFactory.t
@@ -1678,9 +1678,7 @@ $ib = get_annotated_buffer({
 
 is(scalar @{$ib->buffer}, 2, 'minimal - expanded count');
 is($ib->buffer->[0]->allele_string, 'C/T', 'minimal - expanded first allele string');
-# print("Before rejoin\n");
-# use Data::Dumper;
-# print(Dumper($ib->buffer));
+
 $of->rejoin_variants_in_InputBuffer($ib);
 
 is(scalar @{$ib->buffer}, 1, 'minimal - rejoined count');

From 89c3a7d731c39c9e8567b89ab228259b7589768e Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Wed, 18 Dec 2024 08:55:23 +0000
Subject: [PATCH 13/14] Fix tests

---
 t/Parser_Region.t | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/t/Parser_Region.t b/t/Parser_Region.t
index 04763f892..3efcdde05 100644
--- a/t/Parser_Region.t
+++ b/t/Parser_Region.t
@@ -127,13 +127,9 @@ is_deeply($vf, bless( {
   'strand' => '1',
   'variation_name' => '21:25587759-25587758:1/A',
   'map_weight' => 1,
-  'minimised' => 1,
   'allele_string' => '-/A',
-  'original_allele_string' => '-/A',
   'end' => '25587758',
   'start' => '25587759',
-  'original_start' => '25587759',
-  'original_end' => '25587758',
   'seq_region_end' => '25587758',
   'seq_region_start' => '25587759',
   '_line' => ['21:25587759-25587758:1/A']

From 93c1ce9cf45695012af023f96b980292390def46 Mon Sep 17 00:00:00 2001
From: Likhitha Surapaneni
 <10923198+likhitha-surapaneni@users.noreply.github.com>
Date: Thu, 30 Jan 2025 09:34:00 +0000
Subject: [PATCH 14/14] Fixed minimisation for default input

---
 modules/Bio/EnsEMBL/VEP/InputBuffer.pm    |  6 ++--
 t/AnnotationSource_Cache_Variation.t      | 21 ++++--------
 t/AnnotationSource_Cache_VariationTabix.t | 39 ++++++++---------------
 t/InputBuffer.t                           | 10 ++++--
 4 files changed, 31 insertions(+), 45 deletions(-)

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
index eb93f00ca..64d5ba767 100644
--- a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
+++ b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -193,6 +193,7 @@ sub next {
 
     # new chromosome
     if($prev_chr && $vf->{chr} ne $prev_chr) {
+      $self->{minimal}=1 if $vf->{minimised};
       $self->split_variants() if $self->{minimal};
       return $buffer;
     }
@@ -224,7 +225,7 @@ sub next {
         # we can't push the VF back onto the parser, so add it to $pre_buffer
         # and it will get picked up on the following next() call
         push @$pre_buffer, $vf;
-        
+        $self->{minimal}=1 if $vf->{minimised};
         $self->split_variants() if $self->{minimal};
         $prev_start = 0;
         return $buffer;
@@ -245,6 +246,7 @@ sub next {
           }
         }
         $prev_start = $vf->{start};
+        $self->{minimal}=1 if $vf->{minimised};
       }
     }
   }
@@ -257,7 +259,7 @@ sub next {
     die($error_msg);
   }
 
-  $self->split_variants() if $self->{minimal}  ;
+  $self->split_variants() if $self->{minimal};
   return $buffer;
 }
 
diff --git a/t/AnnotationSource_Cache_Variation.t b/t/AnnotationSource_Cache_Variation.t
index a9bd639cf..ed5a32122 100644
--- a/t/AnnotationSource_Cache_Variation.t
+++ b/t/AnnotationSource_Cache_Variation.t
@@ -414,11 +414,11 @@ is_deeply(
 $ib = get_ib([qw(21 8987004 . TA C,TAGCG . . .)]);
 $c->annotate_InputBuffer($ib);
 is_deeply(
-  $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+  $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
   [
     {
-      'a_index' => 1,
-      'a_allele' => 'TAGCG',
+      'a_index' => 0,
+      'a_allele' => 'GCG',
       'b_allele' => 'GCG',
       'b_index' => 0
     }
@@ -444,24 +444,15 @@ is_deeply(
 $ib = get_ib([qw(21 8987004 . TAT TAGCGT,TAGTGT . . .)]);
 $c->annotate_InputBuffer($ib);
 is_deeply(
-  $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+  $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
   [
     {
       'a_index' => 0,
-      'a_allele' => 'AGCGT',
-      'b_allele' => 'GCG',
-      'b_index' => 0
-    },
-    {
-      'a_index' => 1,
-      'a_allele' => 'AGTGT',
+      'a_allele' => 'GTG',
       'b_allele' => 'GTG',
       'b_index' => 1
     }
-  ],
-  'nastiness 4'
-);
-
+  ],'nastiness 4' );
 
 # test old_maf setting
 $p = Bio::EnsEMBL::VEP::Parser::VCF->new({config => $cfg, file => $test_cfg->{test_vcf}, valid_chromosomes => [21]});
diff --git a/t/AnnotationSource_Cache_VariationTabix.t b/t/AnnotationSource_Cache_VariationTabix.t
index 8c04263d7..16f10f16f 100644
--- a/t/AnnotationSource_Cache_VariationTabix.t
+++ b/t/AnnotationSource_Cache_VariationTabix.t
@@ -445,11 +445,11 @@ SKIP: {
   $ib = get_ib([qw(21 8987004 . TA C,TAGCG . . .)]);
   $c->annotate_InputBuffer($ib);
   is_deeply(
-    $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+    $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
     [
       {
-        'a_index' => 1,
-        'a_allele' => 'TAGCG',
+        'a_index' => 0,
+        'a_allele' => 'GCG',
         'b_allele' => 'GCG',
         'b_index' => 0
       }
@@ -475,17 +475,11 @@ SKIP: {
   $ib = get_ib([qw(21 8987004 . TAT TAGCGT,TAGTGT . . .)]);
   $c->annotate_InputBuffer($ib);
   is_deeply(
-    $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+    $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
     [
       {
         'a_index' => 0,
-        'a_allele' => 'AGCGT',
-        'b_allele' => 'GCG',
-        'b_index' => 0
-      },
-      {
-        'a_index' => 1,
-        'a_allele' => 'AGTGT',
+        'a_allele' => 'GTG',
         'b_allele' => 'GTG',
         'b_index' => 1
       }
@@ -494,16 +488,17 @@ SKIP: {
   );
 
   # Test frequency match for 'matched alleles'
-  my $ib_freq = get_ib([qw(21 25005812 . CA CAAA,CAAA . . .)]);
+  my $ib_freq = get_ib([qw(21 25005811 . CA CAAA,CAAA . . .)]);
   $c->{freq_pop} = '1KG_AMR';
   $c->annotate_InputBuffer($ib_freq);
   $vf = $ib_freq->buffer->[0];
   $c->get_frequency_data($vf);
+  
   is_deeply(
     $vf->{_freq_check_freqs},
     {
       '1KG_AMR' => {
-        'AAA' => '0.4107'
+        'AA' => '0.4107'
       },
     },
     'get_frequency_data - matched alleles'
@@ -606,11 +601,11 @@ SKIP: {
   $ib = get_ib([qw(21 8987004 . TA C,TAGCG . . .)]);
   $c->annotate_InputBuffer($ib);
   is_deeply(
-    $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+    $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
     [
       {
-        'a_index' => 1,
-        'a_allele' => 'TAGCG',
+        'a_index' => 0,
+        'a_allele' => 'GCG',
         'b_allele' => 'GCG',
         'b_index' => 0
       }
@@ -635,18 +630,12 @@ SKIP: {
 
   $ib = get_ib([qw(21 8987004 . TAT TAGCGT,TAGTGT . . .)]);
   $c->annotate_InputBuffer($ib);
-  is_deeply(
-    $ib->buffer->[0]->{existing}->[0]->{matched_alleles},
+    is_deeply(
+    $ib->buffer->[1]->{existing}->[0]->{matched_alleles},
     [
       {
         'a_index' => 0,
-        'a_allele' => 'AGCGT',
-        'b_allele' => 'GCG',
-        'b_index' => 0
-      },
-      {
-        'a_index' => 1,
-        'a_allele' => 'AGTGT',
+        'a_allele' => 'GTG',
         'b_allele' => 'GTG',
         'b_index' => 1
       }
diff --git a/t/InputBuffer.t b/t/InputBuffer.t
index 47fba3220..4d5c1a9d4 100644
--- a/t/InputBuffer.t
+++ b/t/InputBuffer.t
@@ -478,13 +478,17 @@ is_deeply(
     'variation_name' => '.',
     'map_weight' => 1,
     'minimised' => 1,
-    'allele_string' => 'CAG/TAG/T',
+    'allele_string' => 'C/T',
     'original_allele_string' => 'CAG/TAG/T',
     'nontrimmed_allele_string' => 'CAG/TAG/T',
-    'end' => 3,
+    'alt_allele' => 'T',
+    'end' => 1,
     'start' => 1,
     'seq_region_start' => 1,
-    'seq_region_end' => 3,
+    'seq_region_end' => 1,
+    'original_start' => 1,
+    'original_end' => 3,
+    'first' => 1
   }, 'Bio::EnsEMBL::Variation::VariationFeature' ),
   'minimal - doesnt affect non-minimisable'
 );