Refine subsampling option and functions

CMakeLists.txt

@@ -168,7 +168,7 @@ include( "${CMAKE_CURRENT_LIST_DIR}/tools/cmake/DownloadDependency.cmake" )
 # These are replaced by tools/cmake/update_dependencies.sh to the hashes that are currently checked out.
 # Thus, do not replace the hashes manually!
 SET( CLI11_COMMIT_HASH "5cb3efabce007c3a0230e4cc2e27da491c646b6c" ) #CLI11_COMMIT_HASH#
-SET( genesis_COMMIT_HASH "91b2221df7921f080d362bbca134708fd962c864" ) #genesis_COMMIT_HASH#
+SET( genesis_COMMIT_HASH "5aeac184fc0b5d78b72e944309a8d688d85fc581" ) #genesis_COMMIT_HASH#
 
 # Call the github download function, which takes four arguments:
 #  - LIBPATH    : Path to the libracy dir where dependencies are stored.

src/options/variant_filter_region.cpp

@@ -30,6 +30,7 @@
 #include "genesis/population/formats/genome_region_reader.hpp"
 #include "genesis/population/formats/gff_reader.hpp"
 #include "genesis/population/formats/map_bim_reader.hpp"
+#include "genesis/population/formats/vcf_common.hpp"
 #include "genesis/population/formats/vcf_input_stream.hpp"
 #include "genesis/population/functions/filter_transform.hpp"
 #include "genesis/population/functions/functions.hpp"

src/options/variant_transform_subsample.cpp

@@ -48,16 +48,21 @@ void VariantTransformSubsampleOptions::add_subsample_opts_to_app(
     );
 
     // Rename samples option.
+    // See https://www.kofler.or.at/bioinformatic/wp-content/uploads/2018/07/pooledAnalysis_part1.pdf
     max_coverage_.option = sub->add_option(
         "--subsample-max-coverage",
         max_coverage_.value,
         "If provided, the nucleotide counts of each sample are subsampled so that they do not "
-        "exceed this given maximum total coverage (sum of the four nucleotide counts). "
+        "exceed this given maximum total coverage (sum of the four nucleotide counts, as well as "
+        "the any `N` and deleted `D` counts). "
         "If they are below this value anyway, they are not changed. "
         "This transformation is useful to limit the maximum coverage. For instance, the diversity "
         "estimators for Theta Pi and Theta Watterson have terms that depend on coverage. "
         "In particular when merging samples such as with `--sample-group-merge-table-file`, "
-        "having an upper limit can hence avoid long compute times."
+        "having an upper limit can hence avoid long compute times. "
+        "Furthermore, a very low Tajima's D, usually indicative of a selective sweep, may be found "
+        "as an artifact in highly covered regions, as such regions have just more sequencing errors. "
+        "To avoid these kinds of biases we recommend to subsample to an uniform coverage. "
         // "This transformation is applied after any filters, so that, e.g., filters high coverage "
         // "remove any unwanted positions first. See `--subsample-method` for the subsampling method."
     );
@@ -110,21 +115,21 @@ void VariantTransformSubsampleOptions::add_subsample_transformation(
     if( method == "subscale" ) {
         variant_input.add_combined_filter_and_transforms(
             [ max_coverage ]( Variant& variant ){
-                transform_subscale( variant, max_coverage );
+                subscale_counts( variant, max_coverage );
                 return true;
             }
         );
     } else if( method == "subsample-with-replacement" ) {
         variant_input.add_combined_filter_and_transforms(
             [ max_coverage ]( Variant& variant ){
-                transform_subsample_with_replacement( variant, max_coverage );
+                subsample_counts_with_replacement( variant, max_coverage );
                 return true;
             }
         );
     } else if( method == "subsample-without-replacement" ) {
         variant_input.add_combined_filter_and_transforms(
             [ max_coverage ]( Variant& variant ){
-                transform_subsample_without_replacement( variant, max_coverage );
+                subsample_counts_without_replacement( variant, max_coverage );
                 return true;
             }
         );