Opened QuickBLAST Git repo

DESCRIPTION

@@ -0,0 +1,16 @@
+Package: QuickBLAST
+Type: Package
+Title: BLAST Files/Sequences Directly From R
+Version: 1.0
+Date: 2023-07-09
+Author: Vishvesh Karthik
+Maintainer: Vishvesh Karthik <vishveshkarthik@gmail.com>
+Description: A BLAST Wrapper to call BLAST directly from R. Depends on ncbi-c++ toolkit (Built and libraries installed)
+License: GPL (>= 2)
+Depends:
+  Rcpp (>= 1.0.10),
+  RcppProgress (>= 0.4.2),
+  arrow (>= 12.0.1.1)
+LinkingTo: Rcpp, RcppProgress
+Encoding: UTF-8
+RoxygenNote: 7.2.3

NAMESPACE

@@ -0,0 +1,3 @@
+useDynLib(QuickBLAST, .registration=TRUE)
+importFrom(Rcpp, evalCpp)
+exportPattern("^[[:alpha:]]+")

R/RcppExports.R

@@ -0,0 +1,85 @@
+# Generated by using Rcpp::compileAttributes() -> do not edit by hand
+# Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393
+
+#' @name CreateNewBLASTInstance
+#'
+#' @title Create new BLAST instance with seq_type, strand, program and BLAST options.
+#'
+#' @description Create a new QuickBLAST C++ object with seq_type, strand, program and BLAST options, which can be used in QuickBLAST::BLAST2Files() and QuickBLAST::BLAST2Seqs()
+#'
+#' @param seq_info (R List) list((int) "seq_type" : 0 - (eNucleotide) (OR) 1 - (eProtein), (int) "strand" = 0 - (ePlus) (OR) 1 - (eMinus), (bool) "save_sequences" = Save Sequences in the arrow::RecordBatch?). Named or unnamed List is accepted but in order.
+#' @param program (string) Name of the BLAST program
+#' @param options (string (or) Named List) List of BLAST options - check COMPLETE::GetAvailableBLASTOptions()
+#' @return C++ Pointer of QuickBLAST Object (Cannot be used in R)
+CreateNewBLASTInstance <- function(seq_info, program, options) {
+    .Call(`_QuickBLAST_CreateNewBLASTInstance`, seq_info, program, options)
+}
+
+#' @name BLAST2Seqs
+#' @title BLAST 2 Sequences Strings
+#'
+#' @description BLAST 2 Sequence Strings and return the RecordBatchVector of BLAST Hits
+#'
+#' @param ptr (QuickBLAST ptr) Pointer to QuickBLAST object created with CreateNewBLASTInstance().
+#' @param query (string) Query Sequence
+#' @param subject (string) Subject Sequence
+#' @return RecordBatchVector of BLAST Hits
+BLAST2Seqs <- function(ptr, query, subject) {
+    .Call(`_QuickBLAST_BLAST2Seqs`, ptr, query, subject)
+}
+
+#' @name BLAST2Folders
+#' @title BLAST 2 Folders
+#'
+#' @description BLAST files of query folder with files of subject folder.
+#'
+#' @note Output file is named as $query-$subject.hits in the outFolder.
+#'
+#' @param ptr (QuickBLAST ptr) Pointer to QuickBLAST object created with CreateNewBLASTInstance().
+#' @param query (std::string) Query FASTA Folder
+#' @param subject (std::string) Subject FASTA Folder
+#' @param extension (std::string) Extension of FASTA files.
+#' @param outFolder (string) Output Folder
+#' @param num_threads (int) Number of Threads
+#' @return List of output filenames
+BLAST2Folders <- function(ptr, query, subject, extension, outFolder, num_threads) {
+    .Call(`_QuickBLAST_BLAST2Folders`, ptr, query, subject, extension, outFolder, num_threads)
+}
+
+#' @name BLAST1Folder
+#' @title BLAST 1 Folder
+#'
+#' @description BLAST the contents of a folder.
+#'
+#' @note Output file is named as $query-$subject.hits in the outFolder.
+#'
+#' @param ptr (QuickBLAST ptr) Pointer to QuickBLAST object created with CreateNewBLASTInstance().
+#' @param input_folder (std::string) Input Folder with FASTA (Query and Subjects)
+#' @param extension (std::string) Extension of FASTA files.
+#' @param outFolder (string) Output Folder
+#' @param num_threads (int) Number of Threads
+#' @return List of output filenames
+BLAST1Folder <- function(ptr, input_folder, extension, outFolder, num_threads) {
+    .Call(`_QuickBLAST_BLAST1Folder`, ptr, input_folder, extension, outFolder, num_threads)
+}
+
+#' @name BLAST2Files
+#' @title BLAST 2 Files
+#'
+#' @description BLAST 2 Files and return the RecordBatchVector of BLAST Hits Asynchronously
+#'
+#' @note Keep return_values_ = FALSE for large files to avoid choking R.
+#'
+#' @param ptr (QuickBLAST ptr) Pointer to QuickBLAST object created with CreateNewBLASTInstance().
+#' @param query (string) Query FASTA File
+#' @param subject (string) Subject FASTA File
+#' @param outFile (string) Output Filename (Arrow Feather/IPC Format)
+#' @param seq_limit (int) Batch Size to BLAST at a time. { -1 = Whole File, 0 - One sequence at a time or > 0 }
+#' @param num_threads (int) Number of Threads
+#' @param show_progress (bool) TRUE - Show progress, Set FALSE for multiple instances
+#' @param return_values (bool) TRUE - Returns values as list, Default - FALSE - Does not return values (Return true on completion)
+#' @return RecordBatchVector of Asynchronous BLAST Hits
+BLAST2Files <- function(ptr, query, subject, outFile, seq_limit, num_threads, show_progress, return_values) {
+    .Call(`_QuickBLAST_BLAST2Files`, ptr, query, subject, outFile, seq_limit, num_threads, show_progress, return_values)
+}
+

src/Makevars

@@ -0,0 +1,3 @@
+CXXSTD = CXX17
+PKG_LIBS = -L/usr/local/lib/ -L/usr/lib/ -larrow -lparquet -lrt -lxncbi -lxobjsimple -lxobjmgr -lblastinput -lblast -lblastdb -lblastdb_format -lxblast -lpthread -lseq -lxser -lgeneral -lseqset -lxutil -lalign_format `$(R_HOME)/bin/Rscript -e "Rcpp:::CxxFlags()"` `$(R_HOME)/bin/Rscript -e "Rcpp:::LdFlags()"` $(SHLIB_OPENMP_CXXFLAGS)
+PKG_CXXFLAGS = -O3 -fPIC -fpermissive -fopenmp -Wextra -shared -Ilibsrc `$(R_HOME)/bin/Rscript -e "Rcpp:::CxxFlags()"` $(SHLIB_OPENMP_CXXFLAGS) -pthread -I/usr/local/src/ -I/usr/local/include/ncbi-tools++/ -I/usr/include/arrow/ -I/usr/include/parquet/ -I/usr/include/ 

src/Makevars.win

@@ -0,0 +1,3 @@
+CXXSTD = CXX17
+PKG_LIBS = -L/usr/local/lib/ -L/usr/lib/ -larrow -lparquet -lrt -lxncbi -lxobjsimple -lxobjmgr -lblastinput -lblast -lblastdb -lblastdb_format -lxblast -lpthread -lseq -lxser -lgeneral -lseqset -lxutil -lalign_format `$(R_HOME)/bin/Rscript -e "Rcpp:::CxxFlags()"` `$(R_HOME)/bin/Rscript -e "Rcpp:::LdFlags()"` $(SHLIB_OPENMP_CXXFLAGS)
+PKG_CXXFLAGS = -O3 -fPIC -fpermissive -fopenmp -Wextra -shared -Ilibsrc `$(R_HOME)/bin/Rscript -e "Rcpp:::CxxFlags()"` $(SHLIB_OPENMP_CXXFLAGS) -pthread -I/usr/local/src/ -I/usr/local/include/ncbi-tools++/ -I/usr/include/arrow/ -I/usr/include/parquet/ -I/usr/include/ 

src/RcppExports.cpp

@@ -0,0 +1,104 @@
+// Generated by using Rcpp::compileAttributes() -> do not edit by hand
+// Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393
+
+#include <Rcpp.h>
+
+using namespace Rcpp;
+
+#ifdef RCPP_USE_GLOBAL_ROSTREAM
+Rcpp::Rostream<true>&  Rcpp::Rcout = Rcpp::Rcpp_cout_get();
+Rcpp::Rostream<false>& Rcpp::Rcerr = Rcpp::Rcpp_cerr_get();
+#endif
+
+// CreateNewBLASTInstance
+RcppExport SEXP CreateNewBLASTInstance(SEXP seq_info, SEXP program, SEXP options);
+RcppExport SEXP _QuickBLAST_CreateNewBLASTInstance(SEXP seq_infoSEXP, SEXP programSEXP, SEXP optionsSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type seq_info(seq_infoSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type program(programSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type options(optionsSEXP);
+    rcpp_result_gen = Rcpp::wrap(CreateNewBLASTInstance(seq_info, program, options));
+    return rcpp_result_gen;
+END_RCPP
+}
+// BLAST2Seqs
+RcppExport SEXP BLAST2Seqs(SEXP ptr, SEXP query, SEXP subject);
+RcppExport SEXP _QuickBLAST_BLAST2Seqs(SEXP ptrSEXP, SEXP querySEXP, SEXP subjectSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type ptr(ptrSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type query(querySEXP);
+    Rcpp::traits::input_parameter< SEXP >::type subject(subjectSEXP);
+    rcpp_result_gen = Rcpp::wrap(BLAST2Seqs(ptr, query, subject));
+    return rcpp_result_gen;
+END_RCPP
+}
+// BLAST2Folders
+RcppExport SEXP BLAST2Folders(SEXP ptr, SEXP query, SEXP subject, SEXP extension, SEXP outFolder, SEXP num_threads);
+RcppExport SEXP _QuickBLAST_BLAST2Folders(SEXP ptrSEXP, SEXP querySEXP, SEXP subjectSEXP, SEXP extensionSEXP, SEXP outFolderSEXP, SEXP num_threadsSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type ptr(ptrSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type query(querySEXP);
+    Rcpp::traits::input_parameter< SEXP >::type subject(subjectSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type extension(extensionSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type outFolder(outFolderSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type num_threads(num_threadsSEXP);
+    rcpp_result_gen = Rcpp::wrap(BLAST2Folders(ptr, query, subject, extension, outFolder, num_threads));
+    return rcpp_result_gen;
+END_RCPP
+}
+// BLAST1Folder
+RcppExport SEXP BLAST1Folder(SEXP ptr, SEXP input_folder, SEXP extension, SEXP outFolder, SEXP num_threads);
+RcppExport SEXP _QuickBLAST_BLAST1Folder(SEXP ptrSEXP, SEXP input_folderSEXP, SEXP extensionSEXP, SEXP outFolderSEXP, SEXP num_threadsSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type ptr(ptrSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type input_folder(input_folderSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type extension(extensionSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type outFolder(outFolderSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type num_threads(num_threadsSEXP);
+    rcpp_result_gen = Rcpp::wrap(BLAST1Folder(ptr, input_folder, extension, outFolder, num_threads));
+    return rcpp_result_gen;
+END_RCPP
+}
+// BLAST2Files
+RcppExport SEXP BLAST2Files(SEXP ptr, SEXP query, SEXP subject, SEXP outFile, SEXP seq_limit, SEXP num_threads, SEXP show_progress, SEXP return_values);
+RcppExport SEXP _QuickBLAST_BLAST2Files(SEXP ptrSEXP, SEXP querySEXP, SEXP subjectSEXP, SEXP outFileSEXP, SEXP seq_limitSEXP, SEXP num_threadsSEXP, SEXP show_progressSEXP, SEXP return_valuesSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type ptr(ptrSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type query(querySEXP);
+    Rcpp::traits::input_parameter< SEXP >::type subject(subjectSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type outFile(outFileSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type seq_limit(seq_limitSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type num_threads(num_threadsSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type show_progress(show_progressSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type return_values(return_valuesSEXP);
+    rcpp_result_gen = Rcpp::wrap(BLAST2Files(ptr, query, subject, outFile, seq_limit, num_threads, show_progress, return_values));
+    return rcpp_result_gen;
+END_RCPP
+}
+
+RcppExport SEXP _rcpp_module_boot_blast_module();
+
+static const R_CallMethodDef CallEntries[] = {
+    {"_QuickBLAST_CreateNewBLASTInstance", (DL_FUNC) &_QuickBLAST_CreateNewBLASTInstance, 3},
+    {"_QuickBLAST_BLAST2Seqs", (DL_FUNC) &_QuickBLAST_BLAST2Seqs, 3},
+    {"_QuickBLAST_BLAST2Folders", (DL_FUNC) &_QuickBLAST_BLAST2Folders, 6},
+    {"_QuickBLAST_BLAST1Folder", (DL_FUNC) &_QuickBLAST_BLAST1Folder, 5},
+    {"_QuickBLAST_BLAST2Files", (DL_FUNC) &_QuickBLAST_BLAST2Files, 8},
+    {"_rcpp_module_boot_blast_module", (DL_FUNC) &_rcpp_module_boot_blast_module, 0},
+    {NULL, NULL, 0}
+};
+
+RcppExport void R_init_QuickBLAST(DllInfo *dll) {
+    R_registerRoutines(dll, NULL, CallEntries, NULL, NULL);
+    R_useDynamicSymbols(dll, FALSE);
+}