start extracting a cellranger-only configuration subset

enclone/Cargo.toml

@@ -38,9 +38,6 @@ string_utils = { path = "../string_utils" }
 vdj_ann = { path = "../vdj_ann" }
 vector_utils = { path = "../vector_utils" }
 
-[target.'cfg(not(windows))'.dependencies]
-pager = "0.16"
-
 [target.'cfg(not(windows))'.dependencies.hdf5]
 git = "https://github.com/10XGenomics/hdf5-rust.git"
 branch = "conda_nov2021"

enclone/src/UNDOC_OPTIONS

@@ -48,16 +48,13 @@ Optional arguments governing input and output files:
 Optional arguments that control printing of individual clonotypes:
 
 - white   = percent of sequences implicated in whitelist expansion.
-- CHAIN_BRIEF: show abbreviated chain column headers
 - DEBUG_TABLE_PRINTING: add print lines to help debug printing of tables.
 - NOTE_SIMPLE: note if the first sequence for the chain is simple, in the sense that it exactly
                equals the concatenation of the right-truncated V with the full J segment.
 
 Other optional arguments:
 
 - FORCE: make joins even if redundant
-- EXP: exploratory code for exact clonotyping on
-- WEAK: for EXP, print all and show weaks
 - GRAPH: show logging from light-heavy graph construction
 - UTR_CON: run experimental UTR consensus code
 - CON_CON: run experimental constant region consensus code
@@ -67,7 +64,6 @@ Other optional arguments:
       2. You want to see the effect of changed annotation code.
 - NPLAIN: reverses PLAIN
 - INDELS: search for and list CDR3s from clonotypes with possible SHM indels (exploratory)
-- NOPRETTY: turn off pretty trace entirely
 - HEAVY_CHAIN_REUSE: look for instances of heavy chain reuse
 - BINARY=filename: generate binary output file
 - PROTO=filename: generate proto output file
@@ -83,13 +79,6 @@ expanded out.
 CELLRANGER: for use if called from cellranger -- changes failure message and prevents exit
             upon normal completion
 
-EXT=filename:
-Given output of an external clonotyping algorithm which took as inputs the pipeline outputs
-for the lenas in enclone.testdata, for each exact subclonotype found by enclone, report its
-composition in the external clonotyping, as clonotype_id[count], ...
-The input file should have lines of the form:
-sample barcode clonotype_id.
-
 SUMMARY_CLEAN: if SUMMARY specified, don't show computational performance stats, so
 we can regress on output
 

enclone/src/graph_filter.rs

@@ -379,7 +379,7 @@ pub fn graph_filter(
                 .insert(tig_bc[i][0].barcode.clone(), BarcodeFate::GraphFilter);
         }
     }
-    if !ctl.gen_opt.ngraph_filter {
+    if !ctl.cr_opt.ngraph_filter {
         erase_if(tig_bc, &to_delete);
     }
     if graph {

enclone/src/info.rs

@@ -222,10 +222,6 @@ pub fn build_info(
                 } else {
                     // maybe can't happen
                     vs.push(rt.clone());
-                    // At one point there was a bug in which the following line was missing.
-                    // This caused a traceback on "enclone 123085 RE".  It is interesting because
-                    // the traceback did not get back to the main program, even with
-                    // "enclone 123085 RE NOPRETTY".
                     vs_notes.push(String::new());
                     vsnx = String::new();
                 }

enclone/src/misc1.rs

@@ -9,8 +9,6 @@ use enclone_core::{
 };
 use equiv::EquivRel;
 use itertools::Itertools;
-#[cfg(not(target_os = "windows"))]
-use pager::Pager;
 
 use std::time::Instant;
 use string_utils::stringme;
@@ -20,44 +18,6 @@ use vector_utils::{
 
 // ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
 
-// This section contains a function that supports paging.  It does not work under Windows, and
-// we describe here all the *known* problems with getting enclone to work under Windows.
-// 1. It does not compile for us.  When we tried, there was a problem with libhdf-5.
-// 2. Paging is turned off, because the pager crate doesn't compile under Windows, and porting
-//    it to Windows appears nontrivial.
-// 3. ANSI escape characters are not handled correctly, at least by default.
-// In addition, we have some concerns about what it would mean to properly test enclone on Windows,
-// given that some users might have older OS installs, and support for ANSI escape characters
-// appears to have been changed in 2018. This is not made easier by the Windows Subsystem for
-// Linux.
-
-#[cfg(not(target_os = "windows"))]
-pub fn setup_pager(pager: bool) {
-    // If the output is going to a terminal, set up paging so that output is in effect piped to
-    // "less -R -F -X -K".
-    //
-    // ∙ The option -R is used to render ANSI escape characters correctly.  We do not use
-    //   -r instead because if you navigate backwards in less -r, stuff gets screwed up,
-    //   which is consistent with the scary stuff in the man page for less at -r.  However -R will
-    //   not display all unicode characters correctly, so those have to be picked carefully,
-    //   by empirically testing that e.g. "echo ◼ | less -R -F -X" renders correctly.
-    //
-    // ∙ The -F option makes less exit immediately if all the output can be seen in one screen.
-    //
-    // ∙ The -X option is needed because we found that in full screen mode on OSX Catalina, output
-    //   was sent to the alternate screen, and hence it appeared that one got no output at all
-    //   from enclone.  This is really bad, so do not turn off this option!
-
-    if pager {
-        Pager::with_pager("less -R -F -X -K").setup();
-    }
-}
-
-#[cfg(target_os = "windows")]
-pub fn setup_pager(_pager: bool) {}
-
-// ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
-
 // Lookup for heavy chain reuse (special purpose experimental option).
 // This is interesting but not likely to yield interesting examples of heavy chain reuse
 // because biologically it doesn't make sense that one would have both H-L1 and H-L2 expanded.

enclone_args/src/lib.rs

@@ -1,8 +1,6 @@
 // Copyright (c) 2021 10x Genomics, Inc. All rights reserved.
 #![allow(clippy::needless_range_loop)]
 
-use io_utils::path_exists;
-
 pub mod load_gex;
 pub mod load_gex_core;
 pub mod load_gex_util;
@@ -14,38 +12,3 @@ pub mod proc_args_post;
 pub mod process_special_arg1;
 pub mod process_special_arg2;
 pub mod read_json;
-
-// parse_csv_pure: same as parse_csv, but don't strip out quotes
-
-pub fn parse_csv_pure(x: &str) -> Vec<&str> {
-    let w = x.char_indices().collect::<Vec<_>>();
-    let mut y = Vec::new();
-    let (mut quotes, mut i) = (0, 0);
-    while i < w.len() {
-        let mut j = i;
-        while j < w.len() {
-            if quotes % 2 == 0 && w[j].1 == ',' {
-                break;
-            }
-            if w[j].1 == '"' {
-                quotes += 1;
-            }
-            j += 1;
-        }
-        let (start, stop) = (w[i].0, w.get(j).map_or(x.len(), |(ind, _)| *ind));
-        y.push(&x[start..stop]);
-        i = j + 1;
-    }
-    if !w.is_empty() && w.last().unwrap().1 == ',' {
-        y.push("");
-    }
-    y
-}
-
-pub fn fnx(outs: &str, name: &str) -> String {
-    let mut file = format!("{outs}/../{name}");
-    if !path_exists(&file) {
-        file = format!("{outs}/{name}");
-    }
-    file
-}

enclone_args/src/load_gex.rs

@@ -19,7 +19,6 @@ use vector_utils::{bin_position, unique_sort};
 pub fn get_gex_info(ctl: &mut EncloneControl) -> Result<GexInfo, String> {
     let mut gex_features = Vec::<Vec<String>>::new();
     let mut gex_barcodes = Vec::<Vec<String>>::new();
-    let mut feature_refs = Vec::<String>::new();
     let mut cluster = Vec::<HashMap<String, usize>>::new();
     let mut cell_type = Vec::<HashMap<String, String>>::new();
     let mut cell_type_specified = Vec::<bool>::new();
@@ -32,12 +31,10 @@ pub fn get_gex_info(ctl: &mut EncloneControl) -> Result<GexInfo, String> {
     let mut h5_paths = Vec::<String>::new();
     let mut feature_metrics = Vec::<HashMap<(String, String), String>>::new();
     let mut json_metrics = Vec::<HashMap<String, f64>>::new();
-    let mut metrics = Vec::<String>::new();
     load_gex(
         ctl,
         &mut gex_features,
         &mut gex_barcodes,
-        &mut feature_refs,
         &mut cluster,
         &mut cell_type,
         &mut cell_type_specified,
@@ -50,7 +47,6 @@ pub fn get_gex_info(ctl: &mut EncloneControl) -> Result<GexInfo, String> {
         &mut h5_paths,
         &mut feature_metrics,
         &mut json_metrics,
-        &mut metrics,
     )?;
     if ctl.gen_opt.gene_scan.is_some() && !ctl.gen_opt.accept_inconsistent {
         let mut allf = gex_features.clone();
@@ -137,7 +133,6 @@ pub fn get_gex_info(ctl: &mut EncloneControl) -> Result<GexInfo, String> {
     Ok(GexInfo {
         gex_features,
         gex_barcodes,
-        feature_refs,
         cluster,
         cell_type,
         cell_type_specified,
@@ -154,6 +149,5 @@ pub fn get_gex_info(ctl: &mut EncloneControl) -> Result<GexInfo, String> {
         have_fb,
         feature_metrics,
         json_metrics,
-        metrics,
     })
 }