From b79e23176b6e50a022c3d05d6dd7c389df9884eb Mon Sep 17 00:00:00 2001
From: Felix Van der Jeugt <felix.vanderjeugt@gmail.com>
Date: Wed, 7 Aug 2019 11:24:21 +0200
Subject: [PATCH] add a minimum frequency argument to jsontree and report

---
 Cargo.toml  | 2 +-
 src/args.rs | 8 ++++++++
 src/main.rs | 8 +++++---
 3 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index b970551..80dbf01 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "umgap"
-version = "0.3.2"
+version = "0.3.3"
 authors = ["Felix Van der Jeugt <felix.vanderjeugt@gmail.com>",
            "Stijn Seghers <stijn.seghers@gmail.com>",
            "Niels De Graef <Niels.DeGraef@UGent.be>",
diff --git a/src/args.rs b/src/args.rs
index 11d4f94..b475343 100644
--- a/src/args.rs
+++ b/src/args.rs
@@ -489,6 +489,10 @@ pub struct JsonTree {
 	#[structopt(short = "r", long = "ranked")]
 	pub ranked_only: bool,
 
+	/// The minimum frequency to be reported
+	#[structopt(short = "f", long = "frequency", default_value = "1")]
+	pub min_frequency: usize,
+
 	/// The NCBI taxonomy tsv-file
 	#[structopt(parse(from_os_str))]
 	pub taxon_file: PathBuf,
@@ -521,6 +525,10 @@ pub struct Report {
 	#[structopt(short = "r", long = "rank", default_value = "species")]
 	pub rank: Rank,
 
+	/// The minimum frequency to be reported
+	#[structopt(short = "f", long = "frequency", default_value = "1")]
+	pub min_frequency: usize,
+
 	/// The NCBI taxonomy tsv-file
 	#[structopt(parse(from_os_str))]
 	pub taxon_file: PathBuf,
diff --git a/src/main.rs b/src/main.rs
index 3eb3b88..9fe39f1 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -534,7 +534,7 @@ fn jsontree(args: args::JsonTree) -> Result<()> {
 	}
 
 	// Recursive json transformation
-	fn to_json(node: &tree::tree::Tree<usize>, aggnode: &tree::tree::Tree<usize>, by_id: &taxon::TaxonList) -> value::Value {
+	fn to_json(node: &tree::tree::Tree<usize>, aggnode: &tree::tree::Tree<usize>, by_id: &taxon::TaxonList, freq: usize) -> value::Value {
 		let root = by_id.get(node.root).unwrap();
 		json!({
 			"name": root.name,
@@ -546,14 +546,15 @@ fn jsontree(args: args::JsonTree) -> Result<()> {
 				"self_count": node.value
 			},
 			"children": node.children.iter().zip(aggnode.children.iter())
-			                         .map(|(n, s)| to_json(n, s, by_id))
+			                         .filter(|&(_n, s)| s.value > freq)
+			                         .map(|(n, s)| to_json(n, s, by_id, freq))
 			                         .collect::<Vec<_>>()
 		})
 	}
 
 	let tree = tree::tree::Tree::new(1, &by_id.ancestry(), &counts)?;
 	let aggtree = tree.aggregate(&ops::Add::add);
-	print!("{}", to_json(&tree, &aggtree, &by_id));
+	print!("{}", to_json(&tree, &aggtree, &by_id, args.min_frequency));
 
 	Ok(())
 }
@@ -660,6 +661,7 @@ fn report(args: args::Report) -> Result<()> {
 	}
 
 	let mut counts = counts.into_iter()
+	                       .filter(|&(_taxon, count)| count > args.min_frequency)
 	                       .map(|(taxon, count)| (count, taxon))
 	                       .collect::<Vec<(taxon::TaxonId, usize)>>();
 	counts.sort();