From 0bf5733046e146109a4fb3ff3b55d9327982928f Mon Sep 17 00:00:00 2001
From: Lucas Czech <luc.czech@gmail.com>
Date: Sat, 8 Jun 2024 16:25:14 +0200
Subject: [PATCH] Add numerical filters and subsampling to sync command

---
 src/commands/convert/sync.cpp | 20 ++++++++++++++++++++
 src/commands/convert/sync.hpp |  6 +++++-
 2 files changed, 25 insertions(+), 1 deletion(-)

diff --git a/src/commands/convert/sync.cpp b/src/commands/convert/sync.cpp
index 24eb8f6..697f438 100644
--- a/src/commands/convert/sync.cpp
+++ b/src/commands/convert/sync.cpp
@@ -49,6 +49,13 @@ void setup_sync( CLI::App& app )
     // Required input of some frequency format (mpileup or vcf at the moment).
     options->variant_input.add_variant_input_opts_to_app( sub );
 
+    // Add the numerical filters
+    options->filter_numerical.add_sample_filter_opts_to_app( sub );
+    options->filter_numerical.add_total_filter_opts_to_app( sub );
+
+    // Also offer subsampling options for this command.
+    options->transform_subsample.add_subsample_opts_to_app( sub );
+
     // -------------------------------------------------------------------------
     //     Settings
     // -------------------------------------------------------------------------
@@ -189,6 +196,19 @@ void run_sync( SyncOptions const& options )
         );
     }
 
+    // Before accessing the variant input, we need to add the filters to it.
+    options.variant_input.add_combined_filter_and_transforms(
+        options.filter_numerical.make_sample_filter()
+    );
+    options.variant_input.add_combined_filter_and_transforms(
+        options.filter_numerical.make_total_filter()
+    );
+
+    // Lastly, apply the subsampling. It is important that this happens after the above numercial
+    // filters above, as otherwise we might subsample to a lower read depth, and then want to apply
+    // a read depth filter, which would not work any more.
+    options.transform_subsample.add_subsample_transformation( options.variant_input );
+
     // If we want to create a gsync file, activate this in the input stream.
     // Needs to be set before any reading access to the stream.
     if( options.gapless.value ) {
diff --git a/src/commands/convert/sync.hpp b/src/commands/convert/sync.hpp
index 558518f..0b51be4 100644
--- a/src/commands/convert/sync.hpp
+++ b/src/commands/convert/sync.hpp
@@ -3,7 +3,7 @@
 
 /*
     grenedalf - Genome Analyses of Differential Allele Frequencies
-    Copyright (C) 2020-2023 Lucas Czech
+    Copyright (C) 2020-2024 Lucas Czech
 
     This program is free software: you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -28,6 +28,8 @@
 
 #include "options/file_output.hpp"
 #include "options/variant_input.hpp"
+#include "options/variant_filter_numerical.hpp"
+#include "options/variant_transform_subsample.hpp"
 
 #include <string>
 #include <vector>
@@ -41,6 +43,8 @@ class SyncOptions
 public:
 
     VariantInputOptions variant_input;
+    VariantFilterNumericalOptions    filter_numerical;
+    VariantTransformSubsampleOptions transform_subsample;
 
     CliOption<bool> without_header  = false;
     CliOption<bool> without_missing = false;