Merge pull request #79 from paulsengroup/improve-hictk-dump

Add --cis-only and --trans-only CLI options to hictk dump. [ci full]
paulsengroup · Dec 11, 2023 · b0bad0b · b0bad0b
2 parents da1fd7d + 1c67c55
commit b0bad0b
Show file tree

Hide file tree

Showing 18 changed files with 617 additions and 258 deletions.
diff --git a/docs/cli_reference.rst b/docs/cli_reference.rst
@@ -139,12 +139,16 @@ hictk dump
                                 Matrix unit (ignored when file is not in .hic format).
     -t,--table TEXT:{chroms,bins,pixels,normalizations,resolutions,cells} [pixels]
                                 Name of the table to dump.
-    -r,--range TEXT [all]  Excludes: --query-file
+    -r,--range TEXT [all]  Excludes: --query-file --cis-only --trans-only
                                 Coordinates of the genomic regions to be dumped following UCSC-style notation (chr1:0-1000).
-    --range2 TEXT [all]  Needs: --range Excludes: --query-file
+    --range2 TEXT [all]  Needs: --range Excludes: --query-file --cis-only --trans-only
                                 Coordinates of the genomic regions to be dumped following UCSC-style notation (chr1:0-1000).
-    --query-file TEXT:(FILE) OR ({-}) Excludes: --range --range2
+    --query-file TEXT:(FILE) OR ({-}) Excludes: --range --range2 --cis-only --trans-only
                                 Path to a BEDPE file with the list of coordinates to be fetched (pass - to read queries from stdin).
+    --cis-only Excludes: --range --range2 --query-file --trans-only
+                                Dump intra-chromosomal interactions only.
+    --trans-only Excludes: --range --range2 --query-file --cis-only
+                                Dump inter-chromosomal interactions only.
     -b,--balance TEXT [NONE]    Balance interactions using the given method.
     --sorted,--unsorted{false}  Return interactions in ascending order.
     --join,--no-join{false}     Output pixels in BG2 format.

diff --git a/docs/reading_interactions.rst b/docs/reading_interactions.rst
@@ -106,3 +106,22 @@ Dump tables other than pixels:
   5000
   10000
   ...
+
+
+Dump cis or trans interactions only:
+
+.. code-block:: console
+
+  user@dev:/tmp$ hictk dump data/4DNFIZ1ZVXC8.hic9 --resolution 1000 --cis-only --join
+
+  chr2L	7000	8000	chr2L	7000	8000	1745
+  chr2L	7000	8000	chr2L	12000	13000	1766
+  chr2L	7000	8000	chr2L	17000	18000	1078
+  ...
+
+  user@dev:/tmp$ hictk dump data/4DNFIZ1ZVXC8.hic9 --resolution 1000 --trans-only --join
+
+  chr2L	7000	8000	chr2R	27000	28000	1
+  chr2L	7000	8000	chr2R	322000	323000	1
+  chr2L	7000	8000	chr2R	397000	398000	1
+  ...
diff --git a/src/hictk/CMakeLists.txt b/src/hictk/CMakeLists.txt
@@ -29,6 +29,9 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/convert/cool_to_hic.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/convert/hic_to_cool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/dump/dump.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/dump/dump_common.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/dump/dump_multiple.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/dump/dump_single.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/fix_mcool/fix_mcool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/load/load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/merge/merge.cpp

diff --git a/src/hictk/cli/cli_dump.cpp b/src/hictk/cli/cli_dump.cpp
@@ -83,6 +83,16 @@ void Cli::make_dump_subcommand() {
       ->check(CLI::ExistingFile | CLI::IsMember({"-"}))
       ->capture_default_str();
 
+  sc.add_flag(
+      "--cis-only",
+      c.cis_only,
+      "Dump intra-chromosomal interactions only.");
+
+  sc.add_flag(
+      "--trans-only",
+      c.trans_only,
+      "Dump inter-chromosomal interactions only.");
+
   sc.add_option(
       "-b,--balance",
       c.normalization,
@@ -111,8 +121,18 @@ void Cli::make_dump_subcommand() {
   // clang-format on
 
   sc.get_option("--range2")->needs(sc.get_option("--range"));
+
   sc.get_option("--query-file")->excludes(sc.get_option("--range"));
   sc.get_option("--query-file")->excludes(sc.get_option("--range2"));
+  sc.get_option("--query-file")->excludes(sc.get_option("--cis-only"));
+  sc.get_option("--query-file")->excludes(sc.get_option("--trans-only"));
+
+  sc.get_option("--cis-only")->excludes(sc.get_option("--trans-only"));
+  sc.get_option("--cis-only")->excludes(sc.get_option("--range"));
+  sc.get_option("--cis-only")->excludes(sc.get_option("--range2"));
+
+  sc.get_option("--trans-only")->excludes(sc.get_option("--range"));
+  sc.get_option("--trans-only")->excludes(sc.get_option("--range2"));
 
   _config = std::monostate{};
 }
@@ -168,6 +188,10 @@ void Cli::validate_dump_subcommand() const {
     errors.emplace_back("--matrix-type=FRAG is not yet supported.");
   }
 
+  if ((c.cis_only || c.trans_only) && c.table != "pixels") {
+    errors.emplace_back("--cis-only and --trans-only require --table=pixels.");
+  }
+
   for (const auto& w : warnings) {
     SPDLOG_WARN(FMT_STRING("{}"), w);
   }