feat: add hidden large PUG threshold cmd line option

src/main.rs

@@ -188,6 +188,17 @@ fn main() -> anyhow::Result<()> {
         ])
         .required(false)
         .hide(true))
+    .arg(arg!(--"large-graph-thresh" <NVERT> "the order (number of nodes) of a PUG above which the alternative resolution strategy will be applied")
+        .default_value_ifs(&[
+            ("resolution", Some("parsimony-gene-em"), Some("1000")),
+            ("resolution", Some("parsimony"), Some("1000")),
+            ("resolution", Some("full"), Some("1000")),
+            ("resolution", Some("parsimony-gene"), Some("1000")),
+            ("resolution", Some("parsimony-gene"), Some("1000")),
+        ])
+        .default_value("0") // for any other mode
+        .required(false)
+        .hide(true))
     .arg(arg!(--"small-thresh" <SMALLTHRESH> "cells with fewer than these many reads will be resolved using a custom approach").default_value("10")
         .hide(true));
 
@@ -398,6 +409,8 @@ fn main() -> anyhow::Result<()> {
         let sa_model: SplicedAmbiguityModel = t.value_of_t("sa-model").unwrap();
         let small_thresh = t.value_of_t("small-thresh").unwrap();
         let filter_list = t.value_of("quant-subset");
+        let large_graph_thresh: usize = t.value_of_t("large-graph-thresh").unwrap();
+        crit!(log, "large graph thresh = {}", large_graph_thresh);
         let umi_edit_dist: u32 = t.value_of_t("umi-edit-dist").unwrap();
         let mut pug_exact_umi = false;
 
@@ -544,6 +557,7 @@ fn main() -> anyhow::Result<()> {
                     pug_exact_umi,
                     sa_model,
                     small_thresh,
+                    large_graph_thresh,
                     filter_list,
                     &cmdline,
                     VERSION,

src/pugutils.rs

@@ -862,6 +862,7 @@ pub fn get_num_molecules(
     tid_to_gid: &[u32],
     gene_eqclass_hash: &mut HashMap<Vec<u32>, u32, ahash::RandomState>,
     hasher_state: &ahash::RandomState,
+    large_graph_thresh: usize,
     log: &slog::Logger,
 ) -> PugResolutionStatistics
 //,)
@@ -919,7 +920,7 @@ pub fn get_num_molecules(
             // are very large.  For components with > 1000 vertices
             // (this should be _very_ rare) we will instead resolve
             // the UMIs in the component using a simpler algorithm.
-            if comp_verts.len() > 1000 {
+            if comp_verts.len() > large_graph_thresh {
                 get_num_molecules_large_component(
                     g,
                     eqmap,

src/quant.rs

@@ -312,6 +312,7 @@ pub fn quantify(
     pug_exact_umi: bool,
     sa_model: SplicedAmbiguityModel,
     small_thresh: usize,
+    large_graph_thresh: usize,
     filter_list: Option<&str>,
     cmdline: &str,
     version: &str,
@@ -354,6 +355,7 @@ pub fn quantify(
             pug_exact_umi,
             sa_model,
             small_thresh,
+            large_graph_thresh,
             filter_list,
             cmdline,
             version,
@@ -384,6 +386,7 @@ pub fn quantify(
             pug_exact_umi,
             sa_model,
             small_thresh,
+            large_graph_thresh,
             filter_list,
             cmdline,
             version,
@@ -410,6 +413,7 @@ pub fn do_quantify<T: Read>(
     pug_exact_umi: bool,
     mut sa_model: SplicedAmbiguityModel,
     small_thresh: usize,
+    large_graph_thresh: usize,
     filter_list: Option<&str>,
     cmdline: &str,
     version: &str,
@@ -924,6 +928,7 @@ pub fn do_quantify<T: Read>(
                                         &tid_to_gid,
                                         &mut gene_eqc,
                                         &s,
+                                        large_graph_thresh,
                                         &log,
                                     );
                                     alt_resolution = pug_stats.used_alternative_strategy; // alt_res;