refactor: untangle data flow structs (#412)

* Extract BarcodeFates type and clean up unnecessary mutability.
* Remove a ton of argument unpacking.
* Remove more unnecessary types that know too much.
* Remove the unused MainEncloneOutput type.
* Remove creation of ref id index into sole place its used.
* Remove pre-computed is_bcr/is_tcr in favor of a method.

enclone/src/graph_filter.rs

@@ -4,12 +4,13 @@
 
 use enclone_core::barcode_fate::BarcodeFate;
 use enclone_core::defs::{EncloneControl, TigData};
+use enclone_core::enclone_structs::BarcodeFates;
 use graph_simple::GraphSimple;
 use io_utils::fwriteln;
 use petgraph::prelude::*;
 use rayon::prelude::*;
 use std::cmp::{max, min};
-use std::collections::HashMap;
+
 use std::io::Write;
 use string_utils::strme;
 use vector_utils::{bin_member, bin_position, erase_if, lower_bound, next_diff12_3, reverse_sort};
@@ -34,7 +35,7 @@ pub fn graph_filter(
     ctl: &EncloneControl,
     tig_bc: &mut Vec<Vec<TigData>>,
     graph: bool,
-    fate: &mut [HashMap<String, BarcodeFate>],
+    fate: &mut [BarcodeFates],
 ) {
     let mut ndels = 0;
     let mut seqs = Vec::<(&[u8], bool, &str, usize)>::new();

enclone/src/info.rs

@@ -3,6 +3,7 @@
 // This file provides the single function build_info.
 
 use enclone_core::barcode_fate::BarcodeFate;
+use enclone_core::enclone_structs::BarcodeFates;
 use vdj_ann::refx;
 
 use self::refx::RefData;
@@ -12,7 +13,7 @@ use debruijn::{dna_string::DnaString, Mer};
 use enclone_core::defs::{CloneInfo, EncloneControl, ExactClonotype};
 use enclone_core::print_tools::emit_codon_color_escape;
 use rayon::prelude::*;
-use std::collections::HashMap;
+
 use std::convert::TryInto;
 use std::fmt::Write;
 use string_utils::strme;
@@ -22,7 +23,7 @@ pub fn build_info(
     refdata: &RefData,
     ctl: &EncloneControl,
     exact_clonotypes: &mut [ExactClonotype],
-    fate: &mut [HashMap<String, BarcodeFate>],
+    fate: &mut [BarcodeFates],
 ) -> Vec<CloneInfo> {
     // Build info about clonotypes.  We create a data structure info.
     // An entry in info is a clonotype having appropriate properties.

enclone/src/join.rs

@@ -29,7 +29,6 @@ use std::io::Write;
 use vector_utils::{bin_member, erase_if, next_diff1_2};
 
 pub fn join_exacts(
-    is_bcr: bool,
     to_bc: &HashMap<(usize, usize), Vec<String>>,
     refdata: &RefData,
     ctl: &EncloneControl,
@@ -104,7 +103,6 @@ pub fn join_exacts(
         // what this means.
 
         join_core(
-            is_bcr,
             i,
             j,
             ctl,

enclone/src/join_core.rs

@@ -9,7 +9,6 @@ use std::collections::HashMap;
 use vdj_ann::refx::RefData;
 
 pub fn join_core<'a>(
-    is_bcr: bool,
     i: usize,
     j: usize,
     ctl: &EncloneControl,
@@ -34,7 +33,6 @@ pub fn join_core<'a>(
                 continue;
             }
             if join_one(
-                is_bcr,
                 k1,
                 k2,
                 ctl,

enclone/src/misc1.rs

@@ -5,13 +5,13 @@
 use enclone_core::{
     barcode_fate::BarcodeFate,
     defs::{CloneInfo, EncloneControl, ExactClonotype, TigData},
+    enclone_structs::BarcodeFates,
 };
 use equiv::EquivRel;
 use itertools::Itertools;
 #[cfg(not(target_os = "windows"))]
 use pager::Pager;
 
-use std::collections::HashMap;
 use std::time::Instant;
 use string_utils::stringme;
 use vector_utils::{
@@ -198,7 +198,7 @@ pub fn lookup_heavy_chain_reuse(
 pub fn cross_filter(
     ctl: &EncloneControl,
     tig_bc: &mut Vec<Vec<TigData>>,
-    fate: &mut [HashMap<String, BarcodeFate>],
+    fate: &mut [BarcodeFates],
 ) {
     // Get the list of dataset origins.  Here we allow the same origin name to have been used
     // for more than one donor, as we haven't explicitly prohibited that.

enclone/src/misc2.rs

@@ -8,10 +8,10 @@ use amino::nucleotide_to_aminoacid_sequence;
 use debruijn::dna_string::DnaString;
 use enclone_core::barcode_fate::BarcodeFate;
 use enclone_core::defs::{EncloneControl, ExactClonotype, Junction, TigData, TigData0, TigData1};
+use enclone_core::enclone_structs::BarcodeFates;
 use io_utils::{fwriteln, open_for_write_new};
 use rayon::prelude::*;
 use std::cmp::{max, min};
-use std::collections::HashMap;
 
 use std::fmt::Write as _;
 use std::io::Write;
@@ -273,7 +273,7 @@ pub fn find_exact_subclonotypes(
     ctl: &EncloneControl,
     tig_bc: &[Vec<TigData>],
     refdata: &RefData,
-    fate: &mut [HashMap<String, BarcodeFate>],
+    fate: &mut [BarcodeFates],
 ) -> Vec<ExactClonotype> {
     let mut exact_clonotypes = Vec::<ExactClonotype>::new();
     let mut r = 0;

enclone_args/src/read_json.rs

@@ -6,6 +6,7 @@ use self::transcript::is_productive_contig;
 use debruijn::dna_string::DnaString;
 use enclone_core::barcode_fate::BarcodeFate;
 use enclone_core::defs::{EncloneControl, OriginInfo, TigData};
+use enclone_core::enclone_structs::BarcodeFates;
 use io_utils::{open_maybe_compressed, path_exists};
 use martian_filetypes::json_file::{Json, LazyJsonReader};
 use martian_filetypes::LazyRead;
@@ -697,20 +698,26 @@ pub struct Annotations {
     pub gex_cells: Vec<Vec<String>>,
     pub gex_cells_specified: Vec<bool>,
     pub tig_bc: Vec<Vec<TigData>>,
-    pub fate: Vec<HashMap<String, BarcodeFate>>,
+    pub fate: Vec<BarcodeFates>,
 }
 
 pub fn parse_json_annotations_files(
     ctl: &EncloneControl,
     refdata: &RefData,
-    to_ref_index: &HashMap<usize, usize>,
 ) -> Result<Annotations, String> {
     // Note: only tracking truncated seq and quals initially
     let ann = if !ctl.gen_opt.cellranger {
         "all_contig_annotations.json"
     } else {
         "contig_annotations.json"
     };
+    let to_ref_index = refdata
+        .id
+        .iter()
+        .take(refdata.refs.len())
+        .enumerate()
+        .map(|(i, &id)| (id as usize, i))
+        .collect();
     let results = ctl
         .origin_info
         .dataset_path
@@ -728,7 +735,7 @@ pub fn parse_json_annotations_files(
                 li,
                 &json,
                 refdata,
-                to_ref_index,
+                &to_ref_index,
                 ctl.gen_opt.reannotate,
                 ctl,
             )

enclone_core/src/defs.rs

@@ -249,6 +249,32 @@ pub struct GeneralOpt {
     pub session_narrative: String,
 }
 
+impl GeneralOpt {
+    pub fn is_bcr(&self) -> bool {
+        if self.tcr {
+            assert!(!self.bcr);
+            return false;
+        }
+        // The original logic for computing this value was based on the confusing
+        // assumption that this is always true if self.tcr is false.
+        // I've preserved this for now, but this line might make sense to add:
+        // assert!(self.bcr);
+        true
+    }
+
+    pub fn is_tcr(&self) -> bool {
+        if self.bcr {
+            assert!(!self.tcr);
+            return false;
+        }
+        // The original logic for computing this value was based on the confusing
+        // assumption that this is always true if self.bcr is false.
+        // I've preserved this for now, but this line might make sense to add:
+        // assert!(self.tcr);
+        true
+    }
+}
+
 #[derive(Clone, PartialEq)]
 pub struct GeneScanOpts {
     pub test: LinearCondition,

enclone_core/src/enclone_structs.rs

@@ -10,44 +10,16 @@ use qd::Double;
 use std::{collections::HashMap, time::Instant};
 use vdj_ann::refx;
 
-#[derive(Clone, Debug, Default)]
-pub struct MainEncloneOutput {
-    pub pics: Vec<String>, // clonotype tables
-    pub last_widths: Vec<u32>,
-    pub svgs: Vec<String>, // SVG objects
-    pub summary: String,   // summary
-    pub metrics: Vec<String>,
-    pub dataset_names: Vec<String>,
-    pub parseable_stdouth: bool,
-    pub noprint: bool,
-    pub noprintx: bool,
-    pub html: bool,
-    pub ngroup: bool,
-    pub pretty: bool,
-}
-
-#[derive(Default)]
-pub struct EncloneState {
-    pub inter: EncloneIntermediates,
-    pub outs: MainEncloneOutput,
-}
-
 #[derive(Default)]
 pub struct EncloneSetup {
     pub ctl: EncloneControl,
     pub ann: String,
     pub gex_info: GexInfo,
     pub tall: Option<Instant>,
     pub refdata: RefData,
-    pub is_bcr: bool,
-    pub to_ref_index: HashMap<usize, usize>,
 }
 
-#[derive(Default)]
-pub struct EncloneIntermediates {
-    pub setup: EncloneSetup,
-    pub ex: EncloneExacts,
-}
+pub type BarcodeFates = HashMap<String, BarcodeFate>;
 
 #[derive(Default, Clone)]
 pub struct EncloneExacts {
@@ -60,8 +32,6 @@ pub struct EncloneExacts {
     pub join_info: Vec<JoinInfo>,
     pub drefs: Vec<DonorReferenceItem>,
     pub sr: Vec<Vec<Double>>,
-    pub fate: Vec<HashMap<String, BarcodeFate>>, // GETS MODIFIED SUBSEQUENTLY
-    pub is_bcr: bool,
     pub allele_data: AlleleData,
 }
 

enclone_core/src/join_one.rs

@@ -64,7 +64,6 @@ pub fn partial_bernoulli_sum(n: usize, k: usize) -> f64 {
 // ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
 
 pub fn join_one<'a>(
-    is_bcr: bool,
     k1: usize,
     k2: usize,
     ctl: &EncloneControl,
@@ -76,6 +75,7 @@ pub fn join_one<'a>(
     refdata: &RefData,
     dref: &[DonorReferenceItem],
 ) -> bool {
+    let is_bcr = ctl.gen_opt.is_bcr();
     // Do not merge onesies or foursies with anything.  Deferred until later.
     // Note that perhaps some foursies should be declared doublets and deleted.
     // Note onesies merging above is turned off so this appears to be moot.

enclone_print/src/define_mat.rs

@@ -77,7 +77,6 @@ pub fn setup_define_mat(orbit: &[i32], info: &[CloneInfo]) -> (Vec<Od>, Vec<usiz
 // algorithm, and is the number of columns (chains) in the clonotype table.
 
 pub fn define_mat(
-    is_bcr: bool,
     to_bc: &HashMap<(usize, usize), Vec<String>>,
     sr: &[Vec<Double>],
     ctl: &EncloneControl,
@@ -171,7 +170,6 @@ pub fn define_mat(
                         if info[l1].lens == info[l2].lens {
                             let mut pot = Vec::<PotentialJoin<'_>>::new();
                             if join_one(
-                                is_bcr,
                                 l1,
                                 l2,
                                 ctl,
@@ -247,7 +245,6 @@ pub fn define_mat(
                 }
                 let mut pot = Vec::<PotentialJoin<'_>>::new();
                 if join_one(
-                    is_bcr,
                     i1,
                     i2,
                     ctl,