Merge pull request #77 from paulsengroup/impl-rename-chroms

Initial implementation of utilities to rename chromosomes in cooler files

.github/workflows/codecov.yml

@@ -216,6 +216,8 @@ jobs:
 
           test/scripts/hictk_merge.sh build/src/hictk/hictk
 
+          test/scripts/hictk_rename_chromosomes.sh build/src/hictk/hictk
+
           test/scripts/hictk_validate.sh build/src/hictk/hictk
 
           test/scripts/hictk_zoomify.sh build/src/hictk/hictk

.github/workflows/macos-ci.yml

@@ -438,6 +438,10 @@ jobs:
         run: |
           test/scripts/hictk_merge.sh bin/hictk
 
+      - name: Test hictk rename-chroms
+        run: |
+          test/scripts/hictk_rename_chromosomes.sh bin/hictk
+
       - name: Test hictk validate
         run: |
           test/scripts/hictk_validate.sh bin/hictk

.github/workflows/ubuntu-ci.yml

@@ -495,6 +495,10 @@ jobs:
         run: |
           test/scripts/hictk_merge.sh bin/hictk
 
+      - name: Test hictk rename-chroms
+        run: |
+          test/scripts/hictk_rename_chromosomes.sh bin/hictk
+
       - name: Test hictk validate
         run: |
           test/scripts/hictk_validate.sh bin/hictk

src/hictk/CMakeLists.txt

@@ -22,6 +22,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_fix_mcool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_merge.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_rename_chromosomes.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_validate.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_zoomify.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/balance/balance.cpp
@@ -35,6 +36,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/fix_mcool/fix_mcool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/load/load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/merge/merge.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/rename_chromosomes/rename_chromosomes.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/validate/validate.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/zoomify/zoomify.cpp)
 

src/hictk/cli/cli.cpp

@@ -37,6 +37,8 @@ auto Cli::parse_arguments() -> Config {
       _subcommand = subcommand::load;
     } else if (_cli.get_subcommand("merge")->parsed()) {
       _subcommand = subcommand::merge;
+    } else if (_cli.get_subcommand("rename-chromosomes")->parsed()) {
+      _subcommand = subcommand::rename_chromosomes;
     } else if (_cli.get_subcommand("validate")->parsed()) {
       _subcommand = subcommand::validate;
     } else if (_cli.get_subcommand("zoomify")->parsed()) {
@@ -85,6 +87,8 @@ std::string_view Cli::subcommand_to_str(subcommand s) noexcept {
       return "load";
     case merge:
       return "merge";
+    case rename_chromosomes:
+      return "rename-chromosomes";
     case validate:
       return "validate";
     case zoomify:
@@ -107,6 +111,7 @@ void Cli::make_cli() {
   make_fix_mcool_subcommand();
   make_load_subcommand();
   make_merge_subcommand();
+  make_rename_chromosomes_subcommand();
   make_validate_subcommand();
   make_zoomify_subcommand();
 }
@@ -131,6 +136,9 @@ void Cli::validate_args() const {
     case merge:
       validate_merge_subcommand();
       break;
+    case rename_chromosomes:
+      validate_rename_chromosomes_subcommand();
+      break;
     case validate:
       break;
     case zoomify:
@@ -161,6 +169,9 @@ void Cli::transform_args() {
     case merge:
       transform_args_merge_subcommand();
       break;
+    case rename_chromosomes:
+      transform_args_rename_chromosomes_subcommand();
+      break;
     case validate:
       break;
     case zoomify:

src/hictk/cli/cli_rename_chromosomes.cpp

@@ -0,0 +1,107 @@
+// Copyright (C) 2023 Roberto Rossini <roberros@uio.no>
+//
+// SPDX-License-Identifier: MIT
+
+#include <fmt/std.h>
+
+#include <CLI/CLI.hpp>
+#include <cassert>
+#include <string>
+
+#include "hictk/tools/cli.hpp"
+#include "hictk/tools/config.hpp"
+
+namespace hictk::tools {
+void Cli::make_rename_chromosomes_subcommand() {
+  auto& sc =
+      *_cli.add_subcommand("rename-chromosomes", "Rename chromosomes found in a Cooler file.")
+           ->fallthrough()
+           ->preparse_callback([this]([[maybe_unused]] std::size_t i) {
+             assert(_config.index() == 0);
+             _config = RenameChromosomesConfig{};
+           });
+
+  _config = RenameChromosomesConfig{};
+  auto& c = std::get<RenameChromosomesConfig>(_config);
+
+  // clang-format off
+  sc.add_option(
+      "uri",
+      c.uri,
+      "Path to a or .[ms]cool file (Cooler URI syntax supported).")
+      ->required();
+
+  sc.add_option(
+      "--name-mappings",
+      c.path_to_name_mappings,
+      "Path to a two column TSV with pairs of chromosomes to be renamed.\n"
+      "The first column should contain the original chromosome name,\n"
+      "while the second column should contain the destination name to use when renaming."
+  );
+
+  sc.add_flag(
+      "--add-chr-prefix",
+      c.add_chr_prefix,
+      "Prefix chromosome names with \"chr\".")
+      ->capture_default_str();
+
+  sc.add_flag(
+      "--remove-chr-prefix",
+      c.remove_chr_prefix,
+      "Remove prefix \"chr\" from chromosome names.")
+      ->capture_default_str();
+
+  sc.add_option(
+      "-v,--verbosity",
+      c.verbosity,
+      "Set verbosity of output to the console.")
+      ->check(CLI::Range(1, 4))
+      ->capture_default_str();
+  // clang-format on
+
+  sc.get_option("--name-mappings")->excludes(sc.get_option("--add-chr-prefix"));
+  sc.get_option("--name-mappings")->excludes(sc.get_option("--remove-chr-prefix"));
+  sc.get_option("--add-chr-prefix")->excludes(sc.get_option("--remove-chr-prefix"));
+  sc.alias("rename-chroms");
+
+  _config = std::monostate{};
+}
+
+void Cli::validate_rename_chromosomes_subcommand() const {
+  assert(_cli.get_subcommand("rename-chromosomes")->parsed());
+
+  const auto& c = std::get<RenameChromosomesConfig>(_config);
+
+  std::vector<std::string> errors;
+
+  if (!cooler::utils::is_cooler(c.uri) && !cooler::utils::is_multires_file(c.uri) &&
+      !cooler::utils::is_scool_file(c.uri)) {
+    errors.emplace_back(
+        fmt::format(FMT_STRING("File \"{}\" does not appear to be a Cooler file."), c.uri));
+  }
+
+  const auto& sc = *_cli.get_subcommand("rename-chromosomes");
+  if (sc.get_option("--name-mappings")->empty() && sc.get_option("--add-chr-prefix")->empty() &&
+      sc.get_option("--remove-chr-prefix")->empty()) {
+    errors.emplace_back(
+        "please specify exactly one of --name-mappings, --add-chr-prefix, --remove-chr-prefix");
+  }
+
+  if (!errors.empty()) {
+    throw std::runtime_error(
+        fmt::format(FMT_STRING("the following error(s) where encountered while validating CLI "
+                               "arguments and input file(s):\n - {}\n"),
+                    fmt::join(errors, "\n - ")));
+  }
+}
+
+void Cli::transform_args_rename_chromosomes_subcommand() {
+  assert(_cli.get_subcommand("rename-chromosomes")->parsed());
+  auto& c = std::get<RenameChromosomesConfig>(_config);
+
+  // in spdlog, high numbers correspond to low log levels
+  assert(c.verbosity > 0 && c.verbosity < 5);
+  c.verbosity = static_cast<std::uint8_t>(spdlog::level::critical) - c.verbosity;
+}
+
+}  // namespace hictk::tools

src/hictk/include/hictk/tools/cli.hpp

@@ -198,6 +198,7 @@ class Cli {
     fix_mcool,
     load,
     merge,
+    rename_chromosomes,
     validate,
     zoomify,
   };
@@ -224,6 +225,7 @@ class Cli {
   void make_fix_mcool_subcommand();
   void make_load_subcommand();
   void make_merge_subcommand();
+  void make_rename_chromosomes_subcommand();
   void make_validate_subcommand();
   void make_zoomify_subcommand();
   void make_cli();
@@ -234,6 +236,7 @@ class Cli {
   void validate_fix_mcool_subcommand() const;
   void validate_load_subcommand() const;
   void validate_merge_subcommand() const;
+  void validate_rename_chromosomes_subcommand() const;
   void validate_zoomify_subcommand() const;
   void validate_args() const;
 
@@ -243,6 +246,7 @@ class Cli {
   void transform_args_fix_mcool_subcommand();
   void transform_args_load_subcommand();
   void transform_args_merge_subcommand();
+  void transform_args_rename_chromosomes_subcommand();
   void transform_args_zoomify_subcommand();
   void transform_args();
 };

src/hictk/include/hictk/tools/config.hpp

@@ -133,6 +133,14 @@ struct MergeConfig {
   std::uint8_t verbosity{4};
 };
 
+struct RenameChromosomesConfig {
+  std::string uri{};
+  std::filesystem::path path_to_name_mappings{};
+  bool add_chr_prefix{false};
+  bool remove_chr_prefix{false};
+  std::uint8_t verbosity{4};
+};
+
 struct ValidateConfig {
   std::string uri{};
   bool validate_index{false};
@@ -160,6 +168,7 @@ using Config = std::variant<std::monostate,
                             FixMcoolConfig,
                             LoadConfig,
                             MergeConfig,
+                            RenameChromosomesConfig,
                             ValidateConfig,
                             ZoomifyConfig>;
 // clang-format on

src/hictk/include/hictk/tools/tools.hpp

@@ -14,6 +14,7 @@ namespace hictk::tools {
 [[nodiscard]] int fix_mcool_subcmd(const FixMcoolConfig& c);
 [[nodiscard]] int load_subcmd(const LoadConfig& c);
 [[nodiscard]] int merge_subcmd(const MergeConfig& c);
+[[nodiscard]] int rename_chromosomes_subcmd(const RenameChromosomesConfig& c);
 [[nodiscard]] int validate_subcmd(const ValidateConfig& c);
 [[nodiscard]] int zoomify_subcmd(const ZoomifyConfig& c);
 

src/hictk/main.cpp

@@ -111,6 +111,8 @@ int main(int argc, char** argv) noexcept {
         return load_subcmd(std::get<LoadConfig>(config));
       case sc::merge:
         return merge_subcmd(std::get<MergeConfig>(config));
+      case sc::rename_chromosomes:
+        return rename_chromosomes_subcmd(std::get<RenameChromosomesConfig>(config));
       case sc::validate:
         return validate_subcmd(std::get<ValidateConfig>(config));
       case sc::zoomify: