Update hictk merge to support coolers with variable bin size

paulsengroup · Dec 8, 2023 · 8017d56 · 8017d56
1 parent a003682
commit 8017d56
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 4 deletions.
diff --git a/src/libhictk/bin_table/include/hictk/impl/bin_table_impl.hpp b/src/libhictk/bin_table/include/hictk/impl/bin_table_impl.hpp
@@ -172,8 +172,13 @@ inline std::uint64_t BinTable::map_to_bin_id(std::uint32_t chrom_id, std::uint32
 inline bool BinTable::operator==(const BinTable &other) const {
   return std::visit(
       [&](const auto &t1) {
-        const auto &t2 = std::get<remove_cvref_t<decltype(t1)>>(other._table);
-        return t1 == t2;
+        try {
+          using BinTableT = remove_cvref_t<decltype(t1)>;
+          const auto &t2 = std::get<BinTableT>(other._table);
+          return t1 == t2;
+        } catch (const std::bad_variant_access &) {
+          return false;
+        }
       },
       _table);
 }

diff --git a/src/libhictk/cooler/include/hictk/cooler/impl/utils_merge_impl.hpp b/src/libhictk/cooler/include/hictk/cooler/impl/utils_merge_impl.hpp
@@ -39,13 +39,21 @@ template <typename N>
 inline void validate_bin_size(const std::vector<LightCooler<N>>& coolers) {
   assert(coolers.size() > 1);
   const auto& clr1 = coolers.front();
+  const auto bin_table1 = cooler::File(clr1.uri).bins();
 
   for (std::size_t i = 1; i < coolers.size(); ++i) {
     const auto& clr2 = coolers[i];
-    if (clr1.bin_size != clr2.bin_size) {
+
+    if (clr1.bin_size == 0 || clr2.bin_size == 0) {  // Table(s) with variable bin size
+      const auto bin_table2 = cooler::File(clr2.uri).bins();
+      if (bin_table1 != bin_table2) {
+        throw std::runtime_error(fmt::format(
+            FMT_STRING("cooler \"{}\" and \"{}\" have different bin tables"), clr1.uri, clr2.uri));
+      }
+    } else if (clr1.bin_size != clr2.bin_size) {
       throw std::runtime_error(fmt::format(
           FMT_STRING(
-              "cooler \"{}\" and \"{}\" have different resolutions ({}  and {} respectively)"),
+              "cooler \"{}\" and \"{}\" have different resolutions ({} and {} respectively)"),
           clr1.uri, clr2.uri, clr1.bin_size, clr2.bin_size));
     }
   }

diff --git a/test/units/bin_table/bin_table_test.cpp b/test/units/bin_table/bin_table_test.cpp
@@ -100,6 +100,13 @@ TEST_CASE("BinTable (fixed bins)", "[bin-table][short]") {
     CHECK_THROWS(table.get<BinTableVariable<>>());
   }
 
+  SECTION("operator==") {
+    CHECK(BinTable(table.chromosomes(), 10) == BinTable(table.chromosomes(), 10));
+    CHECK(BinTable(table.chromosomes(), 10) != BinTable(table.chromosomes(), 20));
+    CHECK(BinTable(Reference{table.chromosomes().begin(), table.chromosomes().end() - 1}, 10) !=
+          BinTable(table.chromosomes(), 10));
+  }
+
   SECTION("iterators") {
     const auto& chr1 = table.chromosomes().at("chr1");
     const auto& chr2 = table.chromosomes().at("chr2");
@@ -325,6 +332,23 @@ TEST_CASE("BinTable (variable bins)", "[bin-table][short]") {
     }
   }
 
+  SECTION("operator==") {
+    CHECK(BinTable(table.chromosomes(), start_pos, end_pos) ==
+          BinTable(table.chromosomes(), start_pos, end_pos));
+
+    const std::vector<std::uint32_t> start_pos1{0, 0};
+    const std::vector<std::uint32_t> end_pos1{32, 32};
+    CHECK(BinTable(table.chromosomes(), start_pos, end_pos) !=
+          BinTable(table.chromosomes(), start_pos1, end_pos1));
+
+    const std::vector<std::uint32_t> start_pos2{0};
+    const std::vector<std::uint32_t> end_pos2{32};
+    CHECK(BinTable(Reference{table.chromosomes().begin(), table.chromosomes().end() - 1},
+                   start_pos2, end_pos2) != BinTable(table.chromosomes(), start_pos, end_pos));
+
+    CHECK(BinTable(table.chromosomes(), start_pos, end_pos) != BinTable(table.chromosomes(), 10));
+  }
+
   SECTION("iterators") {
     const auto& chr1 = table.chromosomes().at("chr1");
     const auto& chr2 = table.chromosomes().at("chr2");