Merge pull request #45 from smithlabcode/optimizing_duplicate_remover

Optimizing duplicate remover

Author: andrewdavidsmith

Makefile.am

@@ -1,6 +1,6 @@
 # This file is part of dnmtools
 #
-# Copyright (C) 2022 Andrew D. Smith
+# Copyright (C) 2023 Andrew D. Smith
 #
 # Authors: Andrew D. Smith
 #
@@ -116,7 +116,7 @@ dnmtools_SOURCES += src/analysis/hypermr.cpp
 
 dnmtools_SOURCES += src/utils/clean-hairpins.cpp
 dnmtools_SOURCES += src/utils/guessprotocol.cpp
-dnmtools_SOURCES += src/utils/duplicate-remover.cpp
+dnmtools_SOURCES += src/utils/uniq.cpp
 dnmtools_SOURCES += src/utils/merge-bsrate.cpp
 dnmtools_SOURCES += src/utils/format-reads.cpp
 dnmtools_SOURCES += src/utils/lc-approx.cpp

pipeline/Snakefile

@@ -180,7 +180,7 @@ rule samtools_sort:
 
 rule remove_duplicates:
     """
-    Uses the duplicate-remover tool to remove duplicate reads in a
+    Uses the uniq tool to remove duplicate reads in a
     mapped read file and produce statistics on duplication level.
     """
     input:

src/Makefile

@@ -64,7 +64,7 @@ analysis/levels.o analysis/methcounts.o analysis/methentropy.o \
 analysis/methstates.o analysis/multimethstat.o analysis/pmd.o \
 analysis/roimethstat.o mlml/mlml.o radmeth/dmr.o radmeth/methdiff.o \
 radmeth/radmeth-adjust.o radmeth/radmeth-merge.o radmeth/radmeth.o \
-utils/clean-hairpins.o utils/duplicate-remover.o utils/fast-liftover.o \
+utils/clean-hairpins.o utils/uniq.o utils/fast-liftover.o \
 utils/format-reads.o utils/guessprotocol.o utils/lc-approx.o \
 utils/lift-filter.o utils/merge-bsrate.o utils/merge-methcounts.o \
 utils/symmetric-cpgs.o utils/selectsites.o

src/dnmtools.cpp

@@ -1,7 +1,6 @@
-/* Copyright (C) 2022 University of Southern California and
- *                    Andrew D. Smith and Guilherme Sena
+/* Copyright (C) 2022-2023 Andrew D. Smith and Guilherme Sena
  *
- * Authors: Andrew D. Smith and Song Qiang and Guilherme Sena
+ * Authors: Andrew D. Smith and Guilherme Sena
  *
  * This program is free software: you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as
@@ -53,7 +52,7 @@ int main_radmeth_adjust(int argc, const char **argv);
 int main_radmeth(int argc, const char **argv);
 int main_radmeth_merge(int argc, const char **argv);
 int main_clean_hairpins(int argc, const char **argv);
-int main_duplicate_remover(int argc, const char **argv);
+int main_uniq(int argc, const char **argv);
 int main_fast_liftover(int argc, const char **argv);
 int main_format_reads(int argc, const char **argv);
 int main_guessprotocol(int argc, const char **argv);
@@ -151,7 +150,7 @@ main(int argc, const char **argv) {
   else if (strcmp(argv[1], "radadjust") == 0) ret = main_radmeth_adjust(argc - 1, argv + 1);
   else if (strcmp(argv[1], "radmerge") == 0) ret = main_radmeth_merge(argc - 1, argv + 1);
   else if (strcmp(argv[1], "cleanhp") == 0) ret = main_clean_hairpins(argc - 1, argv + 1);
-  else if (strcmp(argv[1], "uniq") == 0) ret = main_duplicate_remover(argc - 1, argv + 1);
+  else if (strcmp(argv[1], "uniq") == 0) ret = main_uniq(argc - 1, argv + 1);
   else if (strcmp(argv[1], "fastlift") == 0) ret = main_fast_liftover(argc - 1, argv + 1);
   else if (strcmp(argv[1], "format") == 0) ret = main_format_reads(argc - 1, argv + 1);
   else if (strcmp(argv[1], "guessprotocol") == 0) ret = main_guessprotocol(argc - 1, argv + 1);