Fix Issue ultimatesource#45 - Multiple familes in ped file

- Merge family ID and individual ID in `ped.h` to make id unique. Use an "unique/unusual/obscure" delimiter, `_F@I_`, to concatenate these two IDs, then everything will be unique in `table_`. The `id()` function take care of the rest. - A simple delimiter (`_`, `@` ...etc) is avoid to prevent the case that family name is something like `*f1@` or `*f1_`. - Update related unittests.
stevenhwu · Oct 4, 2016 · 1caba64 · 1caba64
1 parent 8a5b064
commit 1caba64
Show file tree

Hide file tree

Showing 5 changed files with 36 additions and 29 deletions.
diff --git a/Tests/DngCall/BamTest.cmake.in b/Tests/DngCall/BamTest.cmake.in
@@ -21,7 +21,7 @@ set(TagLB-CMD @DNG_CALL_EXE@ --rgtag "LB" -f sample-5.3_ref.fasta.gz -p ceu.ped
 set(TagLB-WD "@TESTDATA_DIR@/sample_5_3/")
 set(TagLB-RESULT 0)
 set(TagLB-STDOUT
-  "FORMAT\tGL-1\tGL-2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
+  "FORMAT\tGL-1_F@I_1\tGL-1_F@I_2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
 )
 
 ###############################################################################
@@ -31,7 +31,7 @@ set(TagSM-CMD @DNG_CALL_EXE@ --rgtag "SM" -f sample-5.3_ref.fasta.gz -p ceu.ped
 set(TagSM-WD "@TESTDATA_DIR@/sample_5_3/")
 set(TagSM-RESULT 0)
 set(TagSM-STDOUT
-  "FORMAT\tGL-1\tGL-2\tLB-NA12878\tLB-NA12891\tLB-NA12892"
+  "FORMAT\tGL-1_F@I_1\tGL-1_F@I_2\tLB-NA12878\tLB-NA12891\tLB-NA12892"
 )
 
 ###############################################################################
@@ -41,7 +41,7 @@ set(TagID-CMD @DNG_CALL_EXE@ --rgtag "ID" -f sample-5.3_ref.fasta.gz -p ceu.ped
 set(TagID-WD "@TESTDATA_DIR@/sample_5_3/")
 set(TagID-RESULT 0)
 set(TagID-STDOUT
-  "FORMAT\tGL-1\tGL-2\tSM-NA12891\tSM-NA12892\tSM-NA12878\tLB-NA12878-H06HD.1\tLB-NA12878-H06HD.2\tLB-NA12878-H06JU\\.1\tLB-NA12891-H03N7\\.1\tLB-NA12891-H03N7\\.2\tLB-NA12891-H05F1\\.2\tLB-NA12892-H06JH\\.1\tLB-NA12892-H06JH\\.2\tLB-NA12892-H06JU\\.2"
+  "FORMAT\tGL-1_F@I_1\tGL-1_F@I_2\tSM-NA12891\tSM-NA12892\tSM-NA12878\tLB-NA12878-H06HD.1\tLB-NA12878-H06HD.2\tLB-NA12878-H06JU\\.1\tLB-NA12891-H03N7\\.1\tLB-NA12891-H03N7\\.2\tLB-NA12891-H05F1\\.2\tLB-NA12892-H06JH\\.1\tLB-NA12892-H06JH\\.2\tLB-NA12892-H06JU\\.2"
 )
 
 ###############################################################################
@@ -51,14 +51,14 @@ set(SepHeader1-CMD @DNG_CALL_EXE@ -f sample-5.3_ref.fasta.gz -p ceu.ped -m 0.001
 set(SepHeader1-WD "@TESTDATA_DIR@/sep_header/")
 set(SepHeader1-RESULT 0)
 set(SepHeader1-STDOUT
-  "FORMAT\tGL-1\tGL-2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
+  "FORMAT\tGL-1_F@I_1\tGL-1_F@I_2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
 )
 
 set(SepHeader2-CMD @DNG_CALL_EXE@ -f sample-5.3_ref.fasta.gz -p ceu.ped -m 0.001 -h test1_hdr.sam test1_nohdr.sam)
 set(SepHeader2-WD "@TESTDATA_DIR@/sep_header/")
 set(SepHeader2-RESULT 0)
 set(SepHeader2-STDOUT
-  "FORMAT\tGL-1\tGL-2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
+  "FORMAT\tGL-1_F@I_1\tGL-1_F@I_2\tLB-NA12878-Solexa-135852\tLB-NA12891-Solexa-135851\tLB-NA12892-Solexa-135853"
 )
 
 ###############################################################################

diff --git a/Tests/Unit/dng/relationship_graph_m12.cc b/Tests/Unit/dng/relationship_graph_m12.cc
@@ -133,11 +133,11 @@ BOOST_FIXTURE_TEST_CASE(test_constructor, FixturePedigreeMid ) {
     auto labels = relationship_graph.labels();
 
     const std::vector<std::string> expected_labels = {
-            "GL-1", "GL-2",
-            "GL-4", "GL-5",
-            "GL-9", "GL-10",
-            "GL-3", "GL-6",
-            "GL-11", "GL-8",
+            "GL-1_F@I_1", "GL-1_F@I_2",
+            "GL-1_F@I_4", "GL-1_F@I_5",
+            "GL-1_F@I_9", "GL-1_F@I_10",
+            "GL-1_F@I_3", "GL-1_F@I_6",
+            "GL-1_F@I_11", "GL-1_F@I_8",
             "LB-NA12001:Solexa-001", "LB-NA12002:Solexa-002",
             "LB-NA12003:Solexa-003", "LB-NA12004:Solexa-004",
             "LB-NA12005:Solexa-005", "LB-NA12006:Solexa-006",
@@ -427,12 +427,12 @@ BOOST_FIXTURE_TEST_CASE(test_add_lib_from_rgs, FixturePedigreeMid) {
 
     std::vector<std::string> expected_vertex{
         "GL-unknown",
-        "GL-1", "GL-2",
-        "GL-4", "GL-5",
-        "GL-9", "GL-10",
-        "GL-3", "GL-6", "GL-11",
-        "GL-7", "GL-8",
-        "GL-12",
+        "GL-1_F@I_1", "GL-1_F@I_2",
+        "GL-1_F@I_4", "GL-1_F@I_5",
+        "GL-1_F@I_9", "GL-1_F@I_10",
+        "GL-1_F@I_3", "GL-1_F@I_6", "GL-1_F@I_11",
+        "GL-1_F@I_7", "GL-1_F@I_8",
+        "GL-1_F@I_12",
 
         "SM-NA12001", "SM-NA12002",
         "SM-NA12004", "SM-NA12005",
@@ -654,11 +654,11 @@ BOOST_FIXTURE_TEST_CASE(test_update_labels_node_ids, FixturePedigreeMid) {
     };
 
     std::vector<std::string> expected_labels {
-        "GL-1", "GL-2",
-        "GL-4", "GL-5",
-        "GL-9", "GL-10",
-        "GL-3", "GL-6", "GL-11",
-        "GL-8",
+        "GL-1_F@I_1", "GL-1_F@I_2",
+        "GL-1_F@I_4", "GL-1_F@I_5",
+        "GL-1_F@I_9", "GL-1_F@I_10",
+        "GL-1_F@I_3", "GL-1_F@I_6", "GL-1_F@I_11",
+        "GL-1_F@I_8",
 
         "LB-NA12001:Solexa-001", "LB-NA12002:Solexa-002",
         "LB-NA12003:Solexa-003", "LB-NA12004:Solexa-004",

diff --git a/Tests/Unit/dng/relationship_graph_trio.cc b/Tests/Unit/dng/relationship_graph_trio.cc
@@ -69,8 +69,8 @@ BOOST_FIXTURE_TEST_CASE(test_constructor, FixturePedigree ) {
     auto labels = relationship_graph.labels();
 
     const std::vector<std::string> expected_labels = {
-        "GL-1", // founder 1
-        "GL-2", // founder 2
+        "GL-1_F@I_1", // founder 1
+        "GL-1_F@I_2", // founder 2
         "LB-NA12878:Solexa-135852",  // lib 1
         "LB-NA12891:Solexa-135851",  // lib 2
         "LB-NA12892:Solexa-135853"   // lib 3
@@ -239,9 +239,9 @@ BOOST_FIXTURE_TEST_CASE(test_add_lib_from_rgs, ReadTrioFromFile) {
 
     std::vector<std::string> expected_vertex{
         "GL-unknown",
-        "GL-1",
-        "GL-2",
-        "GL-3",
+        "GL-1_F@I_1",
+        "GL-1_F@I_2",
+        "GL-1_F@I_3",
         "SM-NA12891",
         "SM-NA12892",
         "SM-NA12878",
@@ -376,8 +376,8 @@ BOOST_FIXTURE_TEST_CASE(test_update_labels_node_ids, ReadTrioFromFile) {
             S_MAX,S_MAX,S_MAX,S_MAX,
             2, 3, 4};
     std::vector<std::string> expected_labels {
-        "GL-1",
-        "GL-2",
+        "GL-1_F@I_1",
+        "GL-1_F@I_2",
         "LB-NA12878:Solexa-135852",
         "LB-NA12891:Solexa-135851",
         "LB-NA12892:Solexa-135853"

diff --git a/src/include/dng/io/ped.h b/src/include/dng/io/ped.h
@@ -1,6 +1,8 @@
 /*
  * Copyright (c) 2014-2015 Reed A. Cartwright
+ * Copyright (c) 2016 Steven H. Wu
  * Authors:  Reed A. Cartwright <[email protected]>
+ *           Steven H. Wu <[email protected]>
  *
  * This file is part of DeNovoGear.
  *
@@ -40,6 +42,8 @@ namespace dng {
 namespace io {
 
 class Pedigree {
+    const std::string FAM_IND_DELIM = "_F@I_";
+
 public:
     typedef boost::multi_index_container<std::string,
             boost::multi_index::indexed_by<
@@ -127,6 +131,9 @@ class Pedigree {
         map<string, size_t> child_names;
         child_names.emplace("", 0);
         for(k = 1; k < string_table.size(); ++k) {
+            string_table[k][1] = string_table[k][0] + FAM_IND_DELIM + string_table[k][1];
+            string_table[k][2] = string_table[k][0] + FAM_IND_DELIM + string_table[k][2];
+            string_table[k][3] = string_table[k][0] + FAM_IND_DELIM + string_table[k][3];
             bool success = child_names.emplace(string_table[k][1], k).second;
             // If child name is duplicate, erase it
             if(!success) {

diff --git a/src/include/dng/relationship_graph.h b/src/include/dng/relationship_graph.h
@@ -35,7 +35,7 @@
 #include <dng/peeling.h>
 #include <dng/detail/unit_test.h>
 
-#define DEBUG_RGRAPH 1
+//#define DEBUG_RGRAPH 1
 
 namespace dng {