Introduce a type alias for Stats. Return them from row_fill.

10XGenomics · Mar 30, 2024 · 59da289 · 59da289
1 parent f8af31a
commit 59da289
Show file tree

Hide file tree

Showing 10 changed files with 38 additions and 28 deletions.
diff --git a/enclone_print/src/finish_table.rs b/enclone_print/src/finish_table.rs
@@ -4,6 +4,7 @@ use crate::build_table_stuff::build_table_stuff;
 use crate::print_utils1::make_table;
 use crate::print_utils3::{add_header_text, insert_reference_rows};
 use crate::print_utils5::{build_diff_row, insert_consensus_row};
+use crate::Stat;
 use enclone_core::defs::{justification, ColInfo, EncloneControl, ExactClonotype};
 use enclone_proto::types::DonorReferenceItem;
 use std::collections::HashMap;
@@ -33,7 +34,7 @@ pub fn finish_table(
     dref: &[DonorReferenceItem],
     peer_groups: &[Vec<(usize, u8, u32)>],
     mlog: &mut Vec<u8>,
-    stats: &[(String, Vec<String>)],
+    stats: &[Stat],
     sr: Vec<Sr>,
     extra_args: &[String],
     pcols_sort: &[String],

diff --git a/enclone_print/src/gene_scan.rs b/enclone_print/src/gene_scan.rs
@@ -2,6 +2,8 @@
 
 use enclone_core::{defs::GeneScanOpts, linear_condition::LinearCondition};
 
+use crate::Stat;
+
 pub struct InSet {
     pub test: bool,
     pub control: bool,
@@ -10,8 +12,8 @@ pub struct InSet {
 pub fn gene_scan_test(
     opts: &GeneScanOpts,
     exact: bool,
-    stats: &[(String, Vec<String>)],
-    stats_orig: &[(String, Vec<String>)],
+    stats: &[Stat],
+    stats_orig: &[Stat],
     nexacts: usize,
     n: usize,
 ) -> Vec<InSet> {

diff --git a/enclone_print/src/lib.rs b/enclone_print/src/lib.rs
@@ -15,3 +15,5 @@ pub mod print_utils5;
 pub mod proc_cvar_auto;
 pub mod proc_lvar2;
 pub mod proc_lvar_auto;
+
+pub type Stat = (String, Vec<String>);
diff --git a/enclone_print/src/print_clonotypes.rs b/enclone_print/src/print_clonotypes.rs
@@ -17,6 +17,7 @@ use crate::print_utils3::{
 };
 use crate::print_utils4::{build_show_aa, compute_bu, compute_some_stats, SomeStats};
 use crate::print_utils5::{delete_weaks, vars_and_shares};
+use crate::Stat;
 use enclone_args::proc_args_check::involves_gex_fb;
 use enclone_core::allowed_vars::{CVARS_ALLOWED, CVARS_ALLOWED_PCELL, LVARS_ALLOWED};
 use enclone_core::defs::{CloneInfo, ColInfo};
@@ -247,7 +248,7 @@ pub fn print_clonotypes(
         // There are two passes.  On the first pass we only identify the exact subclonotypes that
         // are junk.  On the second pass we remove those and then print the orbit.
 
-        let mut stats_pass1 = Vec::<Vec<(String, Vec<String>)>>::new();
+        let mut stats_pass1 = Vec::<Vec<Stat>>::new();
         let mut bads = vec![false; exacts.len()];
         for _ in [0] {
             // pass 1
@@ -502,7 +503,7 @@ pub fn print_clonotypes(
 
                 // Set up to record stats that assign a value to each cell for a given variable.
 
-                let mut stats = Vec::<(String, Vec<String>)>::new();
+                let mut stats = Vec::<Stat>::new();
 
                 // Precompute for near and far.
 
@@ -533,13 +534,12 @@ pub fn print_clonotypes(
                     }
                     let clonotype_id = exacts[u];
                     let ex = &exact_clonotypes[clonotype_id];
-                    let mut these_stats = Vec::<(String, Vec<String>)>::new();
 
                     if ctl.clono_filt_opt_def.whitef && !has_whitelist_contamination(ex) {
                         bads[u] = true;
                     }
 
-                    row_fill(
+                    let (_, mut these_stats) = row_fill(
                         1,
                         u,
                         ctl,
@@ -574,7 +574,6 @@ pub fn print_clonotypes(
                         &mut row,
                         &mut res.out_data,
                         &mut cx,
-                        &mut these_stats,
                     )?;
                     stats.append(&mut these_stats.clone());
 
@@ -603,7 +602,7 @@ pub fn print_clonotypes(
 
                 stats.sort_by(|a, b| a.0.cmp(&b.0));
                 stats = {
-                    let mut stats2 = Vec::<(String, Vec<String>)>::new();
+                    let mut stats2 = Vec::<Stat>::new();
                     let mut i = 0;
                     while i < stats.len() {
                         let mut j = i + 1;
@@ -989,7 +988,7 @@ pub fn print_clonotypes(
 
         // Set up to record stats that assign a value to each cell for a given variable.
 
-        let mut stats = Vec::<(String, Vec<String>)>::new();
+        let mut stats = Vec::<Stat>::new();
 
         // Compute some stats;
 
@@ -1033,8 +1032,7 @@ pub fn print_clonotypes(
             }
             let clonotype_id = exacts[u];
             let ex = &exact_clonotypes[clonotype_id];
-            let mut these_stats = Vec::<(String, Vec<String>)>::new();
-            let RowFillGexData { d_all, ind_all } = row_fill(
+            let (RowFillGexData { d_all, ind_all }, mut these_stats) = row_fill(
                 2,
                 u,
                 ctl,
@@ -1069,7 +1067,6 @@ pub fn print_clonotypes(
                 &mut row,
                 &mut res.out_data,
                 &mut cx,
-                &mut these_stats,
             )?;
             stats.append(&mut these_stats.clone());
 
@@ -1137,7 +1134,7 @@ pub fn print_clonotypes(
 
         stats.sort_by(|a, b| a.0.cmp(&b.0));
         let stats_orig = stats.clone();
-        let mut stats2 = Vec::<(String, Vec<String>)>::new();
+        let mut stats2 = Vec::<Stat>::new();
         let mut i = 0;
         while i < stats.len() {
             let mut j = i + 1;

diff --git a/enclone_print/src/print_utils2.rs b/enclone_print/src/print_utils2.rs
@@ -11,6 +11,7 @@ use crate::proc_lvar2::{proc_lvar2_outs, proc_lvar2_stats};
 use crate::proc_lvar_auto::{
     proc_lvar_auto_read_only, proc_lvar_auto_res_outs, proc_lvar_auto_res_stats,
 };
+use crate::Stat;
 use amino::{codon_to_aa, nucleotide_to_aminoacid_sequence};
 use enclone_core::allowed_vars::LVARS_ALLOWED;
 use enclone_core::defs::{AlleleData, ColInfo, EncloneControl, ExactClonotype, GexInfo, POUT_SEP};
@@ -67,7 +68,7 @@ pub fn row_fill(
     dref: &[DonorReferenceItem],
     groups: &HashMap<usize, Vec<usize>>,
     gex_readers: &[Option<GexReaders<'_>>],
-    stats_pass1: &[Vec<(String, Vec<String>)>],
+    stats_pass1: &[Vec<Stat>],
     vdj_cells: &[Vec<String>],
     n_vdj_gex: &[usize],
     lvars: &[String],
@@ -83,8 +84,7 @@ pub fn row_fill(
     row: &mut Vec<String>, // row of human-readable output; only written in pass #2
     out_data: &mut [HashMap<String, String>], // row of parseable output
     cx: &mut [Vec<String>],
-    stats: &mut Vec<(String, Vec<String>)>,
-) -> Result<RowFillGexData, String> {
+) -> Result<(RowFillGexData, Vec<Stat>), String> {
     // Redefine some things to reduce dependencies.
 
     let mat = &rsi.mat;
@@ -221,6 +221,8 @@ pub fn row_fill(
         };
     }
 
+    let mut stats = Vec::<Stat>::new();
+
     'lvar_loop: for (i, &x) in all_lvars.iter().enumerate() {
         // Process VAR_DEF variables.
         if pass == 2 {
@@ -329,12 +331,14 @@ pub fn row_fill(
                 gex_readers,
                 &alt_bcs,
             ) {
-                proc_lvar_auto_res_stats(x, ex, &res, stats);
+                proc_lvar_auto_res_stats(x, ex, &res, &mut stats);
                 if pass == 2 {
                     proc_lvar_auto_res_outs(i, x, u, ctl, extra_args, lvars, &res, row, out_data);
                 }
             } else {
-                proc_lvar2_stats(i, x, ctl, ex, gex_info, &gex_data, lvars, gex_mean, stats);
+                proc_lvar2_stats(
+                    i, x, ctl, ex, gex_info, &gex_data, lvars, gex_mean, &mut stats,
+                );
                 if pass == 2 {
                     proc_lvar2_outs(
                         i,
@@ -567,7 +571,7 @@ pub fn row_fill(
                 varmat,
                 allele_data,
             )? {
-                proc_cvar_auto_res_stats(pass, ex, col, ctl, extra_args, &res, stats);
+                proc_cvar_auto_res_stats(pass, ex, col, ctl, extra_args, &res, &mut stats);
                 if pass == 2 {
                     proc_cvar_auto_res_outs(
                         jj,
@@ -625,7 +629,7 @@ pub fn row_fill(
             }
         }
     }
-    Ok(gex_data)
+    Ok((gex_data, stats))
 }
 
 struct RowFillGexInfo {

diff --git a/enclone_print/src/print_utils4.rs b/enclone_print/src/print_utils4.rs
@@ -13,7 +13,7 @@ use string_utils::TextUtils;
 use vdj_ann::refx::RefData;
 use vector_utils::{bin_member, bin_position, bin_position1_2, unique_sort};
 
-use crate::finish_table::Sr;
+use crate::{finish_table::Sr, Stat};
 
 // ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
 
@@ -516,7 +516,7 @@ pub fn compute_bu(
     d_all: &[Vec<u32>],
     ind_all: &[Vec<u32>],
     mat: &[Vec<Option<usize>>],
-    these_stats: &[(String, Vec<String>)],
+    these_stats: &[Stat],
     refdata: &RefData,
 ) {
     // Very bad computation because of embedded binary search.

diff --git a/enclone_print/src/proc_cvar_auto.rs b/enclone_print/src/proc_cvar_auto.rs
@@ -7,6 +7,7 @@ use crate::print_utils1::{
 };
 use crate::print_utils3::comp_edit;
 use crate::proc_lvar_auto::ProcVarAutoResult;
+use crate::Stat;
 use amino::{codon_to_aa, nucleotide_to_aminoacid_sequence};
 use enclone_core::align_to_vdj_ref::{align_to_vdj_ref, cigar};
 use enclone_core::defs::{AlleleData, ColInfo, EncloneControl, ExactClonotype, POUT_SEP};
@@ -1543,7 +1544,7 @@ pub fn proc_cvar_auto_res_stats(
     ctl: &EncloneControl,
     extra_args: &[String],
     res: &ProcVarAutoResult,
-    stats: &mut Vec<(String, Vec<String>)>,
+    stats: &mut Vec<Stat>,
 ) {
     let ProcVarAutoResult {
         abbr,

diff --git a/enclone_print/src/proc_lvar2.rs b/enclone_print/src/proc_lvar2.rs
@@ -5,6 +5,7 @@
 
 use crate::print_utils2::RowFillGexData;
 use crate::print_utils4::get_gex_matrix_entry;
+use crate::Stat;
 use enclone_core::defs::{EncloneControl, ExactClonotype, GexInfo, POUT_SEP};
 use enclone_core::median::rounded_median;
 use itertools::Itertools;
@@ -133,7 +134,7 @@ pub fn proc_lvar2_stats(
     gex_data: &RowFillGexData,
     lvars: &[String],
     gex_mean: f64,
-    stats: &mut Vec<(String, Vec<String>)>,
+    stats: &mut Vec<Stat>,
 ) {
     // Proceed.
 

diff --git a/enclone_print/src/proc_lvar_auto.rs b/enclone_print/src/proc_lvar_auto.rs
@@ -15,6 +15,8 @@ use string_utils::{abbrev_list, strme, TextUtils};
 use vdj_ann::refx::RefData;
 use vector_utils::{bin_member, bin_position, unique_sort};
 
+use crate::Stat;
+
 pub struct ProcVarAutoResult {
     pub abbr: String,
     pub exact: String,
@@ -1433,7 +1435,7 @@ pub fn proc_lvar_auto_res_stats(
     var: &str,
     ex: &ExactClonotype,
     res: &ProcVarAutoResult,
-    stats: &mut Vec<(String, Vec<String>)>,
+    stats: &mut Vec<Stat>,
 ) {
     let ProcVarAutoResult {
         abbr,

diff --git a/enclone_vars/src/export_code.rs b/enclone_vars/src/export_code.rs
@@ -547,7 +547,7 @@ pub fn export_code(level: usize) -> Vec<(String, String)> {
             cx: &mut Vec<Vec<String>>,
             varmat: &Vec<Vec<Vec<u8>>>,
             out_data: &mut Vec<HashMap<String, String>>,
-            stats: &mut Vec<(String, Vec<String>)>,
+            stats: &mut Vec<Stat>,
             allele_data: &AlleleData,
         ) -> Result<bool, String> {
 
@@ -709,7 +709,7 @@ pub fn export_code(level: usize) -> Vec<(String, String)> {
             ctl: &EncloneControl,
             extra_args: &Vec<String>,
             out_data: &mut Vec<HashMap<String, String>>,
-            stats: &mut Vec<(String, Vec<String>)>,
+            stats: &mut Vec<Stat>,
             lvars: &Vec<String>,
             row: &mut Vec<String>,
             fate: &Vec<HashMap<String, String>>,