From 1ab7383e751d79c4c34df6d8eb1b94efd3a1d7b9 Mon Sep 17 00:00:00 2001
From: Iago Bonnici <iago.bonnici@umontpellier.fr>
Date: Fri, 6 Sep 2024 19:44:03 +0200
Subject: [PATCH] Export scalar tree data as CSV.
---
Cargo.toml | 1 +
src/bin/aphid/main.rs | 27 +++++--
src/output.rs | 1 +
src/output/csv.rs | 169 ++++++++++++++++++++++++++++++++++++++++++
src/output/detail.rs | 4 +-
5 files changed, 195 insertions(+), 7 deletions(-)
create mode 100644 src/output/csv.rs
diff --git a/Cargo.toml b/Cargo.toml
index c256cf0..692d900 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -29,6 +29,7 @@ snafu = "0.8.2"
serde_json = { version = "1.0.127", features = ["preserve_order"] }
color-print = { git = "https://gitlab.com/iago-lito/color-print", branch = "dev", version = "0.3.6" }
color-print-proc-macro = { git = "https://gitlab.com/iago-lito/color-print", branch = "dev", version = "0.3.6" }
+csv = "1.3.0"
[dev-dependencies]
rand = "0.8.5"
diff --git a/src/bin/aphid/main.rs b/src/bin/aphid/main.rs
index bbeb082..617b7af 100644
--- a/src/bin/aphid/main.rs
+++ b/src/bin/aphid/main.rs
@@ -11,14 +11,14 @@ use aphid::{
interner::{Interner, ResolvedSymbol, SpeciesSymbol},
it_mean::SummedMean,
ln_likelihood, optimize_likelihood,
- output::detail,
+ output::{self, detail},
Config, GeneTree, GeneTriplet, GenesForest, LocalGeneTriplet, VERSION,
};
use clap::Parser;
use color_print::{ceprintln, cformat, cprintln};
use float_eq::float_eq;
use serde_json::json;
-use snafu::{ensure, Snafu};
+use snafu::{ensure, ResultExt, Snafu};
mod args;
@@ -26,6 +26,7 @@ mod args;
mod out {
pub(super) const CONFIG: &str = "config.json";
pub(super) const DETAIL: &str = "detail.json";
+ pub(super) const CSV: &str = "trees.csv";
}
// Run and terminate error bubbling if any.
@@ -91,6 +92,7 @@ fn run() -> Result<(), Error> {
// Output full per-tree detail.
write_detail(&output, &details)?;
+ write_csv(&output, &details)?;
// Fit the model.
learn(&triplets, &config)?;
@@ -248,7 +250,7 @@ fn topology_filter<'i>(
triplet_means_sum += mean_triplet;
rest_means_sum += mean_rest;
- detail.included_topology = true;
+ detail.topology_included = true;
detail.mean_lengths = detail::MeanLengths {
total: Some(mean_length),
triplet: Some(mean_triplet),
@@ -323,7 +325,7 @@ fn geometry_filter(
tree.is_shape_included(mean_triplet, mean_rest, global_shape, max);
detail.local_shape = Some(shape_ratio);
if pass {
- detail.included_geometry = pass;
+ detail.geometry_included = pass;
included.push(i);
} else {
n_excluded += 1;
@@ -384,6 +386,19 @@ fn write_detail(output: &Path, details: &[detail::Tree]) -> Result<(), Error> {
Ok(())
}
+fn write_csv(output: &Path, details: &[detail::Tree]) -> Result<(), Error> {
+ let path = output.join(out::CSV);
+ cprintln!("Summarize scalar values to <b>{}</>.", path.display());
+ let file = File::create(path)?;
+ let mut wtr = csv::Writer::from_writer(file);
+ for detail in details {
+ let record: output::csv::Record = detail.into();
+ wtr.serialize(record).context(CsvErr)?;
+ }
+ wtr.flush()?;
+ Ok(())
+}
+
fn display_summary(included: &[usize], details: &[detail::Tree], config: &Config) {
println!("\nSummary:");
@@ -431,7 +446,7 @@ fn display_summary(included: &[usize], details: &[detail::Tree], config: &Config
if let Some(max) = config.filters.max_clock_ratio {
let n = details
.iter()
- .map(|d| u64::from(!d.included_geometry))
+ .map(|d| u64::from(!d.geometry_included))
.sum();
cprintln!(
" - <g>{n}</> tree shape{} rejected \
@@ -483,4 +498,6 @@ enum Error {
InputConsistency { mess: String },
#[snafu(transparent)]
Learn { source: aphid::learn::Error },
+ #[snafu(display("Could not serialize record to CSV:\n{source}"))]
+ Csv { source: csv::Error },
}
diff --git a/src/output.rs b/src/output.rs
index 585d6fa..87533ae 100644
--- a/src/output.rs
+++ b/src/output.rs
@@ -2,3 +2,4 @@
mod config;
pub mod detail;
+pub mod csv;
diff --git a/src/output/csv.rs b/src/output/csv.rs
new file mode 100644
index 0000000..ef2e012
--- /dev/null
+++ b/src/output/csv.rs
@@ -0,0 +1,169 @@
+// Extract all scalar data from the detailed output
+// and fit it into a large csv table to ease parsing by biologists.
+
+use serde::Serialize;
+
+use super::detail::{Internal, MeanLengths, Outgroup, Top, Tree, Triplet, TripletAnalysis};
+use crate::{
+ gene_tree::{NbBases, TripletTopology},
+ BranchLength,
+};
+
+// One record per tree.
+#[derive(Serialize)]
+#[allow(clippy::struct_excessive_bools)] // On purpose for CSV splatting.
+pub struct Record<'i> {
+ // Global tree information.
+ tree_id: &'i str,
+ n_bases: NbBases,
+ n_nodes_raw: usize,
+ n_nodes_pruned: usize,
+
+ //----------------------------------------------------------------------------------------------
+ // Topology-related data.
+
+ // Triplet section.
+ triplet_n_missing: usize,
+ triplet_lca: Option<usize>,
+ triplet_n_paraphyletic: usize,
+ triplet_included: bool,
+ triplet_topology: Option<TripletTopology>,
+ triplet_branch_length_a: Option<NbBases>,
+ triplet_branch_length_b: Option<NbBases>,
+ triplet_branch_length_c: Option<NbBases>,
+ triplet_branch_length_d: Option<NbBases>,
+ triplet_resolved: Option<bool>,
+
+ // Outgroup section.
+ outgroup_n_missing: usize,
+ outgroup_lca: Option<usize>,
+ outgroup_n_paraphyletic: usize,
+ outgroup_included: bool,
+
+ // Tree top section.
+ top_lca: Option<usize>,
+ top_n_external: Option<usize>,
+ top_n_internal_triplet: Option<usize>,
+ top_n_internal_outgroup: Option<usize>,
+ top_included: Option<bool>,
+
+ topology_included: bool,
+
+ //----------------------------------------------------------------------------------------------
+ // Geometry-related data.
+ mean_lengths_triplet: Option<BranchLength>,
+ mean_lengths_outgroup_other: Option<BranchLength>,
+ mean_lengths_total: Option<BranchLength>,
+
+ local_shape: Option<BranchLength>,
+ geometry_included: bool,
+
+ //----------------------------------------------------------------------------------------------
+ mutation_rate: Option<f64>,
+}
+
+//==================================================================================================
+// Trivial extraction / flattening from the detailed tree information.
+
+impl<'i> From<&Tree<'i>> for Record<'i> {
+ fn from(tree: &Tree<'i>) -> Self {
+ // Destructure all.
+ let &Tree {
+ id: tree_id,
+ n_bases,
+ n_nodes_raw,
+ n_nodes_pruned,
+ triplet:
+ Triplet {
+ lca: triplet_lca,
+ missing: ref triplet_missing,
+ paraphyletic: ref triplet_paraphyletic,
+ analysis: ref triplet_analysis,
+ included: triplet_included,
+ },
+ outgroup:
+ Outgroup {
+ lca: outgroup_lca,
+ missing: ref outgroup_missing,
+ paraphyletic: ref outgroup_paraphyletic,
+ included: outgroup_included,
+ },
+ ref top,
+
+ topology_included,
+ mean_lengths:
+ MeanLengths {
+ total: mean_lengths_total,
+ triplet: mean_lengths_triplet,
+ outgroup_other: mean_lengths_outgroup_other,
+ },
+ local_shape,
+ geometry_included,
+ mutation_rate,
+ } = tree;
+
+ // Deeper into (optional) triplet information destructuring.
+ let (triplet_topology, tl, triplet_resolved) =
+ if let Some(TripletAnalysis { topology, branches_lengths: ref bl, resolved }) =
+ *triplet_analysis
+ {
+ (Some(topology), Some(bl), Some(resolved))
+ } else {
+ (None, None, None)
+ };
+
+ // Deeper into (optional) tree top information destructuring.
+ let (top_lca, top_n_external, int, top_included) =
+ if let Some(Top { lca, ref external, ref internal, included }) = *top {
+ (
+ Some(lca),
+ Some(external.len()),
+ Some(internal),
+ Some(included),
+ )
+ } else {
+ (None, None, None, None)
+ };
+
+ let (top_n_internal_triplet, top_n_internal_outgroup) =
+ if let Some(Some(Internal { triplet, outgroup })) = int {
+ (Some(triplet.len()), Some(outgroup.len()))
+ } else {
+ (None, None)
+ };
+
+ // All is eventually flattened.
+ Record {
+ tree_id,
+ n_bases,
+ n_nodes_raw,
+ n_nodes_pruned,
+ triplet_lca,
+ triplet_n_missing: triplet_missing.len(),
+ triplet_n_paraphyletic: triplet_paraphyletic.len(),
+ triplet_included,
+ triplet_topology,
+ triplet_branch_length_a: tl.map(|l| l.a),
+ triplet_branch_length_b: tl.map(|l| l.b),
+ triplet_branch_length_c: tl.map(|l| l.c),
+ triplet_branch_length_d: tl.map(|l| l.d),
+ triplet_resolved,
+ outgroup_lca,
+ outgroup_n_missing: outgroup_missing.len(),
+ outgroup_n_paraphyletic: outgroup_paraphyletic.len(),
+ outgroup_included,
+ top_lca,
+ top_n_external,
+ top_n_internal_triplet,
+ top_n_internal_outgroup,
+ top_included,
+ topology_included,
+ mean_lengths_triplet,
+ mean_lengths_outgroup_other,
+ mean_lengths_total,
+ local_shape,
+ geometry_included,
+ mutation_rate,
+ }
+ }
+}
diff --git a/src/output/detail.rs b/src/output/detail.rs
index 0228f7b..412ac6e 100644
--- a/src/output/detail.rs
+++ b/src/output/detail.rs
@@ -39,7 +39,7 @@ pub struct Tree<'i> {
pub top: Option<Top<'i>>,
/// Raised if the tree passed the topology filter.
- pub included_topology: bool,
+ pub topology_included: bool,
/// Mean branches lengths,
/// undefined if the none of the species set
@@ -53,7 +53,7 @@ pub struct Tree<'i> {
pub local_shape: Option<f64>,
/// Raised if the tree passed the geometry filter.
- pub included_geometry: bool,
+ pub geometry_included: bool,
/// Estimated mutation rate for this tree.
/// (No estimate if the tree was excluded from analysis.)
--
GitLab