initial upload

CMakeLists.txt

@@ -0,0 +1,91 @@
+# ===========================================================================
+#                  SeqAn - The Library for Sequence Analysis
+# ===========================================================================
+# File: /sandbox/jagla/apps/clean_ngs/CMakeLists.txt
+#
+# CMakeLists.txt file for clean_ngs.
+# ===========================================================================
+
+cmake_minimum_required (VERSION 2.8.2)
+project (seqan_sandbox_jagla_apps_clean_ngs)
+message (STATUS "Configuring sandbox/jagla/apps/clean_ngs")
+
+# ----------------------------------------------------------------------------
+# Dependencies
+# ----------------------------------------------------------------------------
+
+# Search SeqAn and select dependencies.
+set (SEQAN_FIND_DEPENDENCIES ZLIB)
+find_package (SeqAn REQUIRED)
+
+# ----------------------------------------------------------------------------
+# Build Setup
+# ----------------------------------------------------------------------------
+
+# Add include directories.
+include_directories (${SEQAN_INCLUDE_DIRS})
+
+# Add definitions set by find_package (SeqAn).
+add_definitions (${SEQAN_DEFINITIONS})
+
+# Update the list of file names below if you add source files to your application.
+add_executable (clean_ngs
+                clean_ngs.cpp
+                clean_ngs.h
+                adaptor_record.h
+                clean_ngs_app.cpp)
+
+# Add dependencies found by find_package (SeqAn).
+target_link_libraries (clean_ngs ${SEQAN_LIBRARIES})
+
+# Add CXX flags found by find_package (SeqAn).
+set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SEQAN_CXX_FLAGS}")
+
+# ----------------------------------------------------------------------------
+# Installation
+# ----------------------------------------------------------------------------
+
+# Set variables for installing, depending on the selected build type.
+if (NOT SEQAN_PREFIX_SHARE_DOC)
+  seqan_setup_install_vars (clean_ngs)
+endif (NOT SEQAN_PREFIX_SHARE_DOC)
+
+# Install clean_ngs in ${PREFIX}/bin directory
+install (TARGETS clean_ngs
+         DESTINATION bin)
+
+# Install non-binary files for the package to "." for app builds and
+# ${PREFIX}/share/doc/clean_ngs for SeqAn release builds.
+install (FILES LICENSE
+               README
+         DESTINATION ${SEQAN_PREFIX_SHARE_DOC})
+#install (FILES example/fasta1.fa
+#               example/fasta2.fa
+#         DESTINATION ${SEQAN_PREFIX_SHARE_DOC}/example)
+
+# ----------------------------------------------------------------------------
+# App Test
+# ----------------------------------------------------------------------------
+
+seqan_add_app_test (clean_ngs)
+
+# ----------------------------------------------------------------------------
+# Setup Command Tool Description for Generic Workflow Nodes
+# ----------------------------------------------------------------------------
+
+#Include executable pair_align in CTD structure.
+set (SEQAN_CTD_EXECUTABLES ${SEQAN_CTD_EXECUTABLES} clean_ngs CACHE INTERNAL "")
+
+# ----------------------------------------------------------------------------
+# CPack Install
+# ----------------------------------------------------------------------------
+
+if (SEQAN_BUILD_SYSTEM STREQUAL "APP:clean_ngs")
+  set (CPACK_PACKAGE_NAME "clean_ngs")
+  set (CPACK_PACKAGE_DESCRIPTION_SUMMARY "clean_ngs")
+  set (CPACK_DEBIAN_PACKAGE_MAINTAINER "Bernd Jagla <[email protected]>")
+  set (CPACK_PACKAGE_VENDOR "Bernd Jagla <[email protected]>")
+
+  seqan_configure_cpack_app (clean_ngs "clean_ngs")
+endif (SEQAN_BUILD_SYSTEM STREQUAL "APP:clean_ngs")
+

INFO

@@ -0,0 +1,21 @@
+Name: adapterRemoval
+Author: Bernd Jagla <[email protected]>
+Maintainer: Bernd Jagla <[email protected]>
+License: GPL v3
+Copyright: 2008-2012, FU Berlin
+Status: under development
+Description: One-line description for app adapterRemoval.
+ This is the long text description for the app.  You should wrap text to 78
+ characters and you have to wrap it to 80 characters.  The lines must have one
+ leading space.  Successive lines in this format will be word-wrapped when
+ displayed.  Leading space is stripped off.
+ .
+ Use empty lines with one dot to insert an empty line / break paragraphs.
+ .
+  Start lines with two spaces to make them display verbatim.  Lines that
+  cannot be displayed horizontally will be hard-wrapped (not at word bor-
+  ders), so maybe do your own hyphenation.
+ .
+ Do not use tab characters, their effect is not predictable.
+ .
+ The INFO file format is oriented after the Debian controfields format.

TODO

@@ -0,0 +1,36 @@
+Clean NGS -- TODO
+=================
+
+ * By default, print user-friendly output to stderr.  Add an option with the
+   log files for Bernd's statistics scripts.
+
+ * Document adapter file definition in text section. See:
+   http://trac.seqan.de/wiki/Tutorial/ParsingCommandLineArguments
+
+ * Create more comprehensive list of tests.  Should include adapters
+   at the end and in the middle of the read, quality based trimming on
+   each side.
+
+ * Tests with paired end data are important and should be added.
+
+ * Adaptor placement with paired end reads.
+
+ * Windows/Unix
+
+ * Ns are allowed in the adapter sequence.
+
+ * Possibly add option to annotate read id/meta data line with
+   information about leader sequence or adapter.  E.g. (I recommend 1):
+
+   (1) char buffer[1000];
+       snprintf(buffer, 999, " foo bar %d %d", 1, 2);
+       append(record.id1, buffer);
+
+   (2) std::stringstream ss;
+       ss << " foo bar " << 1 << " " << 2;
+       append(record.id1, ss.str());
+       // ss.str(""); ss.clear();
+
+ * Benchmark against old version.
+
+ * Test in KNIME.