Replace is_valid with refactored version

vdj_ann/src/annotate.rs

@@ -3,8 +3,8 @@
 // This file contains code to annotate a contig, in the sense of finding alignments
 // to VDJ reference contigs.  Also to find CDR3 sequences.  And some related things.
 
-use crate::refx::RefData;
 use crate::transcript::is_valid;
+use crate::{refx::RefData, transcript::ContigStatus};
 use align_tools::affine_align;
 use amino::{aa_seq, have_start};
 use bio_edit::alignment::AlignmentOperation::{Del, Ins, Match, Subst, Xclip, Yclip};
@@ -228,16 +228,16 @@
     }
 }
 
 pub fn annotate_seq_core(
     b: &DnaString,
     refdata: &RefData,
     ann: &mut Vec<(i32, i32, i32, i32, i32)>,
     allow_weak: bool,
     allow_improper: bool,
     abut: bool,
     log: &mut Vec<u8>,
     verbose: bool,
 ) {
     // The DNA string representation is inefficient because it stores bases as packed k-mers
     // which requires a lot of array bounds checks when unpacking which was a hot path
     // we found when profiling the CI job.  To avoid those in the inner
@@ -1596,8 +1596,8 @@
             // deletion.  Be very careful to respect this if editing!
 
             let (mut win1, mut win2) = (false, false);
             let c1 = m1 == m2 && !have_utr_align2 && err1_nu == err2_nu && outside1 > outside2;
             let c2 = m2 == m1 && !have_utr_align1 && err2_nu == err1_nu && outside2 > outside1;
             if (zstop1 > zstop2 + 20 && (outside2 <= 10.0 || total2 - share <= 10))
                 || (outside1 >= 10.0 && outside2 <= 1.0 && err1 - err2 <= 2.5)
             {
@@ -1606,8 +1606,8 @@
             } else if (outside2 <= 10.0 || total2 - share <= 10)
                 && (m1 < m2
                     || (m1 == m2 && err1 < err2 && !c2)
                     || (m1 == m2 && err1 == err2 && outside1 > outside2)
                     || (m1 == m2 && err1 == err2 && outside1 == outside2 && t1 < t2)
                     || c1)
             {
                 win1 = true;
@@ -1623,8 +1623,8 @@
             } else if (outside1 <= 10.0 || total1 - share <= 10)
                 && (m2 < m1
                     || (m2 == m1 && err2 < err1 && !c1)
                     || (m2 == m1 && err2 == err1 && outside2 > outside1)
                     || (m2 == m1 && err2 == err1 && outside2 == outside1 && t2 < t1)
                     || c2)
             {
                 win2 = true;
@@ -3018,7 +3018,9 @@
     pub invalidated_umis: Option<Vec<String>>, // invalidated UMIs
     pub is_cell: bool,                       // was the barcode declared a cell?
     pub productive: Option<bool>,            // productive?  (null means not full length)
-    pub filtered: bool,                      // true and never changed (unused field)
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub productive_criteria: Option<ContigStatus>,
+    pub filtered: bool, // true and never changed (unused field)
 
     pub is_gex_cell: Option<bool>, // Was the barcode declared a cell by Gene expression data, if available
     pub is_asm_cell: Option<bool>, // Was the barcode declared a cell by the VDJ assembler
@@ -3050,7 +3052,8 @@
         invalidated_umis: Option<Vec<String>>,   // invalidated UMIs
         is_cellx: bool,                          // was the barcode declared a cell?
         productivex: bool,                       // productive?
-        jsupp: Option<JunctionSupport>,          // num reads, umis supporting junction
+        prod_criteria: ContigStatus, // criteria used to determine if a contig is productive
+        jsupp: Option<JunctionSupport>, // num reads, umis supporting junction
     ) -> ContigAnnotation {
         let mut vstart = -1_i32;
         for i in 0..ann.len() {
@@ -3139,6 +3142,7 @@
             invalidated_umis,
             is_cell: is_cellx,
             productive: Some(productivex),
+            productive_criteria: Some(prod_criteria),
             filtered: true,
             junction_support: jsupp,
             // These need to be populated by the assembler explicitly as needed
@@ -3173,13 +3177,11 @@
         non_validated_umis: Option<Vec<String>>, // non-validated UMIs
         invalidated_umis: Option<Vec<String>>,   // invalidated UMIs
         is_cell: bool,                           // was the barcode declared a cell?
-        is_gd: Option<bool>,                     // is gamma/delta mode
         jsupp: Option<JunctionSupport>,          // num reads, umis supporting junction
     ) -> ContigAnnotation {
         let mut ann = Vec::<(i32, i32, i32, i32, i32)>::new();
         annotate_seq(b, refdata, &mut ann, true, false, true);
-        let mut log = Vec::<u8>::new();
-        let productive = is_valid(b, refdata, &ann, false, &mut log, is_gd);
+        let (productive, contig_status) = is_valid(b, refdata, &ann);
         ContigAnnotation::from_annotate_seq(
             b,
             q,
@@ -3194,6 +3196,7 @@
             invalidated_umis,
             is_cell,
             productive,
+            contig_status,
             jsupp,
         )
     }
@@ -3351,7 +3354,6 @@
             None,
             false, // is_cell, should be changed to None
             None,
-            None,
         );
 
         // println!("{:#?}", annotation);