Use CMAPLE in weekly-tree pipeline, not raxml + garli

.gitignore

@@ -2,3 +2,4 @@ build
 .cache
 .ccls-cache
 __pycache__
+/.vscode

README.md

@@ -17,7 +17,6 @@
 
 - install homebrew https://brew.sh
 - brew install llvm ninja meson libomp cmake brotli zlib xz gnu-time catch2
-- pip3 install mypy
 
 ## Installing dependencies on Ubuntu
 

bin/seqdb-update

@@ -39,8 +39,6 @@ def main(args: argparse.Namespace):
                         # only aligned are stored
                         ae_backend.raw_sequence.calculate_hash(sequence)
                         sequences_by_subtype.setdefault(sequence.type_subtype, []).append(sequence)
-                    else:
-                        print(f"{sequence.name:40s} {sequence.aa}")
                 except Exception as err:
                     print(f"> {err}: {sequence.name}\n    {sequence.aa}")
                 reader.context.messages_from_backend(sequence_messages)

cc/tree/newick.cc

@@ -174,7 +174,7 @@ namespace ae::tree::newick
         {
             static constexpr auto whitespace = dsl::ascii::space / dsl::ascii::newline;
             static constexpr auto rule = dsl::p<internal_node> + dsl::semicolon + dsl::eof;
-            static constexpr auto max_recursion_depth = 1000;
+            static constexpr auto max_recursion_depth = 10000;
 
             static constexpr auto value = lexy::forward<result_t>;
         };

cc/whocc/xlsx/sheet-extractor.cc

@@ -96,7 +96,7 @@ static const std::regex re_VIDRL_serum_name{"^(?:[AB]/)?([A-Z][A-Z ]+)/?([0-9]+)
 static const std::regex re_VIDRL_serum_id{"^(" pattern_VIDRL_serum_id "|" pattern_CRICK_serum_id ")$", regex_icase};
 static const std::regex re_VIDRL_serum_id_with_days{"^[AF][0-9][0-9][0-9][0-9]-[0-9]+D$", regex_icase};
 
-static const std::regex re_human_who_serum{R"(^\s*(.*(HUMAN|WHO|NORMAL)|GOAT|POST? VAX|VAX POOL|SERA POOL)\b)", regex_icase}; // "POST VAX": VIDRL H3 HI 2021, "HUMAN VAX": VIDRL H3 HI 2022
+static const std::regex re_human_who_serum{R"(^\s*(.*(HUMAN|WHO|NORMAL)|GOAT|POST? VAX|VAX POOL)\b)", regex_icase}; // "POST VAX": VIDRL H3 HI 2021, "HUMAN VAX": VIDRL H3 HI 2022
 
 #pragma GCC diagnostic pop
 
@@ -339,16 +339,13 @@ std::string ae::xlsx::v1::Extractor::titer(size_t ag_no, size_t sr_no) const
 void ae::xlsx::v1::Extractor::find_titers(warn_if_not_found winf)
 {
     std::vector<std::pair<nrow_t, range<ncol_t>>> rows;
-    AD_DEBUG("Sheet {} rows:{}", sheet().name(), sheet().number_of_rows());
+    // AD_DEBUG("Sheet {}", sheet().name());
     for (nrow_t row{0}; row < sheet().number_of_rows(); ++row) {
         auto titers = sheet().titer_range(row);
-        AD_DEBUG("Row:{} titers: ({} valid:{}) {}:{}", row, titers.length(), titers.valid(), titers.first, titers.second);
         adjust_titer_range(row, titers);
-        AD_DEBUG("Row:{} titers: ({} valid:{}) {}:{}", row, titers.length(), titers.valid(), titers.first, titers.second);
         if (titers.valid() && titers.length() > 2 && titers.first > ncol_t{0} && valid_titer_row(row, titers))
             rows.emplace_back(row, std::move(titers));
     }
-    AD_DEBUG("Rows with titers: {}", rows.size());
 
     if (!ranges::all_of(rows, [&rows](const auto& en) { return en.second == rows[0].second; })) {
         fmt::memory_buffer report; // fmt::format(rows, "{}", "\n  "));

proj/weekly-tree/collapse_short_branches.R

@@ -0,0 +1,59 @@
+#! /usr/bin/env Rscript
+main = function() {
+    message("COLLAPSE SHORT BRANCH LENGTHS")
+
+    .libPaths(
+        c(
+            "/home/samt123/R/x86_64-pc-linux-gnu-library/4.4",
+            .libPaths()
+        )
+    )
+
+    args <- commandArgs(trailingOnly = TRUE)
+
+    inphy.txt = args[[1]]
+    tol = as.numeric(args[[2]])
+
+    # file names -------------------------------------------------------------
+    inphy = readLines(inphy.txt, n = 1)
+
+    outphy = paste0(
+        paste0(
+            fs::path_ext_remove(inphy),
+            "-col."
+        ),
+        fs::path_ext(inphy)
+    )
+
+    outphy.txt = paste0(
+        paste0(
+            fs::path_ext_remove(inphy.txt),
+            "-col."
+        ),
+        fs::path_ext(inphy.txt)
+    )
+
+    if (fs::file_exists(outphy) & fs::file_exists(outphy.txt)) {
+        message(
+            outphy,
+            " and ",
+            outphy.txt,
+            " already exist! Continuing without re-collapsing branches"
+        )
+        return(0)
+    }
+
+    message("Reading tree")
+    phy = ape::read.tree(file = inphy)
+
+    message("Collapsing branches shorter than ", tol)
+    phy_col = ape::di2multi(phy, tol = tol)
+    message(ape::Nnode(phy), " -> ", ape::Nnode(phy_col))
+
+    message("Writing tree")
+    ape::write.tree(phy_col, file = outphy)
+    message("Writing ", outphy.txt)
+    writeLines(outphy, outphy.txt)
+}
+
+main()