From 1ba6cc1a4a150357753d1316852cbcc009906bff Mon Sep 17 00:00:00 2001
From: Alessandro Vullo <avullo@ebi.ac.uk>
Date: Fri, 2 Aug 2013 14:13:03 +0000
Subject: [PATCH] Changed expected GTF in some cases: the start/stop codons
should not have been there (incomplete CDS annotation)
---
modules/t/circularSlice.t | 22 ++++++++++++++++++++++
modules/t/gtfSerialiser.t | 9 +++++----
2 files changed, 27 insertions(+), 4 deletions(-)
diff --git a/modules/t/circularSlice.t b/modules/t/circularSlice.t
index 129cef8f40..a118d526d0 100644
--- a/modules/t/circularSlice.t
+++ b/modules/t/circularSlice.t
@@ -45,6 +45,28 @@ is($slice->seq_region_length, $SEQ_REGION_LENGTH,
"seq_region_length == $SEQ_REGION_LENGTH");
is($slice->adaptor, $slice_adaptor, "adaptor is adaptor");
+# centrepoint method
+is($slice->centrepoint, 5402639, "slice centre point");
+
+# _split private method
+my ($sl1, $sl2) = $slice->_split;
+isa_ok($sl1, 'Bio::EnsEMBL::CircularSlice');
+is($sl1->is_circular, 1,"subslice is circular");
+is($sl1->seq_region_name, $CHR,"subslice seq region name $CHR");
+is($sl1->start, $START, "subslice start == $START");
+is($sl1->end, $SEQ_REGION_LENGTH, "subslice end == $SEQ_REGION_LENGTH");
+
+isa_ok($sl2, 'Bio::EnsEMBL::CircularSlice');
+is($sl2->is_circular, 1,"subslice is circular");
+is($sl2->seq_region_name, $CHR,"subslice seq region name $CHR");
+is($sl2->start, 1, "subslice start == 1");
+is($sl2->end, $END, "subslice end == $END");
+
+# seq method
+my $seq = $slice->seq;
+is(length $seq, 205278, "sequence length");
+exit;
+
my $slstart = $slice_adaptor->fetch_by_region($COORD_SYSTEM, $CHR, 1, $END);
my $slend = $slice_adaptor->fetch_by_region($COORD_SYSTEM, $CHR, $START,
$SEQ_REGION_LENGTH);
diff --git a/modules/t/gtfSerialiser.t b/modules/t/gtfSerialiser.t
index c66b9b1623..6ed0ede5e7 100644
--- a/modules/t/gtfSerialiser.t
+++ b/modules/t/gtfSerialiser.t
@@ -54,9 +54,10 @@ my $transcripts_gtf =
20\tprotein_coding\texon\t30315002\t30315126\t.\t+\t.\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"5\"; gene_biotype \"protein_coding\"; exon_id \"ENSE00001155819\";
20\tprotein_coding\tCDS\t30315002\t30315126\t.\t+\t1\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"5\"; gene_biotype \"protein_coding\"; protein_id \"ENSP00000252021\";
20\tprotein_coding\texon\t30318805\t30318881\t.\t+\t.\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"6\"; gene_biotype \"protein_coding\"; exon_id \"ENSE00001180793\";
-20\tprotein_coding\tCDS\t30318805\t30318878\t.\t+\t2\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"6\"; gene_biotype \"protein_coding\"; protein_id \"ENSP00000252021\";
-20\tprotein_coding\tstop_codon\t30318879\t30318881\t.\t+\t0\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"6\"; gene_biotype \"protein_coding\";
+20\tprotein_coding\tCDS\t30318805\t30318881\t.\t+\t2\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"6\"; gene_biotype \"protein_coding\"; protein_id \"ENSP00000252021\";
",
+# last line 20\tprotein_coding\tstop_codon\t30318879\t30318881\t.\t+\t0\tgene_id \"ENSG00000174873\"; transcript_id \"ENST00000252021\"; exon_number \"6\"; gene_biotype \"protein_coding\";
+# CDS end on previous line extended to stop_codon length, was 30318878
ENST00000202017 => "20\tprotein_coding\texon\t30327735\t30327869\t.\t-\t.\tgene_id \"ENSG00000088356\"; transcript_id \"ENST00000202017\"; exon_number \"1\"; gene_name \"C20orf126\"; gene_biotype \"protein_coding\"; transcript_name \"C20orf126\"; exon_id \"ENSE00001155739\";
20\tprotein_coding\tCDS\t30327735\t30327869\t.\t-\t0\tgene_id \"ENSG00000088356\"; transcript_id \"ENST00000202017\"; exon_number \"1\"; gene_name \"C20orf126\"; gene_biotype \"protein_coding\"; transcript_name \"C20orf126\"; protein_id \"ENSP00000202017\";
@@ -72,12 +73,11 @@ my $transcripts_gtf =
ENST00000246203 => "20\tprotein_coding\texon\t30565065\t30566129\t.\t-\t.\tgene_id \"ENSG00000125979\"; transcript_id \"ENST00000246203\"; exon_number \"1\"; gene_name \"TSPYL3\"; gene_biotype \"protein_coding\"; transcript_name \"TSPYL3\"; exon_id \"ENSE00000859919\";
20\tprotein_coding\tCDS\t30565065\t30566129\t.\t-\t0\tgene_id \"ENSG00000125979\"; transcript_id \"ENST00000246203\"; exon_number \"1\"; gene_name \"TSPYL3\"; gene_biotype \"protein_coding\"; transcript_name \"TSPYL3\"; protein_id \"ENSP00000246203\";
-20\tprotein_coding\tstart_codon\t30566127\t30566129\t.\t-\t0\tgene_id \"ENSG00000125979\"; transcript_id \"ENST00000246203\"; exon_number \"1\"; gene_name \"TSPYL3\"; gene_biotype \"protein_coding\"; transcript_name \"TSPYL3\";
",
+# last line 20\tprotein_coding\tstart_codon\t30566127\t30566129\t.\t-\t0\tgene_id \"ENSG00000125979\"; transcript_id \"ENST00000246203\"; exon_number \"1\"; gene_name \"TSPYL3\"; gene_biotype \"protein_coding\"; transcript_name \"TSPYL3\";
ENST00000201961 => "20\tprotein_coding\texon\t30911297\t30911383\t.\t-\t.\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"1\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; exon_id \"ENSE00000661216\";
20\tprotein_coding\tCDS\t30911297\t30911383\t.\t-\t0\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"1\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; protein_id \"ENSP00000201961\";
-20\tprotein_coding\tstart_codon\t30911381\t30911383\t.\t-\t0\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"1\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\";
20\tprotein_coding\texon\t30903618\t30903773\t.\t-\t.\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"2\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; exon_id \"ENSE00000661215\";
20\tprotein_coding\tCDS\t30903618\t30903773\t.\t-\t0\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"2\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; protein_id \"ENSP00000201961\";
20\tprotein_coding\texon\t30896671\t30896782\t.\t-\t.\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"3\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; exon_id \"ENSE00000560920\";
@@ -87,6 +87,7 @@ my $transcripts_gtf =
20\tprotein_coding\texon\t30885729\t30885800\t.\t-\t.\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"5\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; exon_id \"ENSE00000661212\";
20\tprotein_coding\tCDS\t30885729\t30885800\t.\t-\t0\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"5\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\"; protein_id \"ENSP00000201961\";
"
+# third line 20\tprotein_coding\tstart_codon\t30911381\t30911383\t.\t-\t0\tgene_id \"ENSG00000088303\"; transcript_id \"ENST00000201961\"; exon_number \"1\"; gene_name \"Q9NQF5\"; gene_biotype \"protein_coding\"; transcript_name \"Q9NQF5\";
};
my $mtdb = Bio::EnsEMBL::Test::MultiTestDB->new();
--
GitLab