Fix: Support for chromosomes >1GB in size

Makefile.simple

@@ -0,0 +1,65 @@
+# Simple Makefile for TRF
+# No autotools required - just plain gcc
+
+CC = gcc
+# Note: -fno-align-loops removed as it's not supported by clang
+CFLAGS = -O2 -Wall -fno-align-functions
+LDFLAGS = -lm
+
+# Platform detection
+UNAME_S := $(shell uname -s)
+UNAME_M := $(shell uname -m)
+
+ifeq ($(UNAME_S),Linux)
+    PLATFORM = linux
+    ifeq ($(UNAME_M),x86_64)
+        TARGET = linux64
+    else
+        TARGET = linux32
+    endif
+endif
+
+ifeq ($(UNAME_S),Darwin)
+    PLATFORM = mac
+    TARGET = mac
+endif
+
+VERSION = 4.10.0
+
+# Source files
+# Note: tr30dat.c is included via trfrun.h, so we only compile trf.c
+SRCDIR = src
+SOURCES = $(SRCDIR)/trf.c
+HEADERS = $(SRCDIR)/tr30dat.h $(SRCDIR)/tr30dat.c $(SRCDIR)/trfrun.h $(SRCDIR)/trfclean.h
+
+# Output binary
+BINARY = trf
+VERSIONED_BINARY = trf$(VERSION).$(TARGET).exe
+
+# Build target
+all: $(BINARY)
+
+$(BINARY): $(SOURCES) $(HEADERS)
+	@echo "Building TRF for $(PLATFORM) ($(TARGET))..."
+	$(CC) $(CFLAGS) -DUNIXCONSOLE $(SOURCES) -o $(BINARY) $(LDFLAGS)
+	@echo "Build complete: $(BINARY)"
+
+# Create versioned binary
+versioned: $(BINARY)
+	cp $(BINARY) $(VERSIONED_BINARY)
+	@echo "Created versioned binary: $(VERSIONED_BINARY)"
+
+# Install (optional)
+install: $(BINARY)
+	install -m 755 $(BINARY) /usr/local/bin/
+
+# Clean
+clean:
+	rm -f $(BINARY) $(VERSIONED_BINARY)
+	@echo "Clean complete"
+
+# Test build
+test: $(BINARY)
+	./$(BINARY) -v
+
+.PHONY: all versioned install clean test

README.md

@@ -61,52 +61,58 @@ To obtain current and/or earlier, pre-compiled versions of TRF:
 
 ## Instructions for Compiling ##
 To compile TRF, you will need a C compiler (e.g., gcc, clang) with the standard library installed.
-We have tested compiling and installing TRF under UNIX-based systems (Linux, macOS) and for Windows under Cygwin/MinGW.
+We have tested compiling and installing TRF under UNIX-based systems (Linux, macOS including Apple Silicon/ARM) and for Windows under Cygwin/MinGW.
 
-Brief instructions (advanced):
+### Simple Build (Recommended) ###
+
+The easiest way to build TRF requires only gcc/clang - no autotools needed:
 
 ```bash
-# Check actual version
-tar xzvf trf-4.10.0.tar.gz
-cd trf-4.10.0
-mkdir build
-cd build
-../configure
-make
-# To install to system
-sudo make install
-# To copy binary elsewhere
-cp src/trf DESTINATION
+# Build with the simple script
+./build.sh
 ```
 
-If you are cloning the repository, replace the first two lines above with:
+That's it! The binary `trf` is ready to use.
+
+Alternatively, you can use the simple Makefile:
 
 ```bash
-git clone https://github.com/Benson-Genomics-Lab/TRF.git
-cd TRF
+make -f Makefile.simple
 ```
 
-Step by step:
+Or compile directly with a single command:
 
-- Open a terminal window and change directory to the TRF directory created by a clone of this repository.
-- Create a directory named 'build': `mkdir build` and change directory to that
-- Type `../configure` in the terminal.
-- Type `make` in the terminal.
-- If you wish to install the binary, type `sudo make install` in the terminal. Otherwise, simply copy the binary from the `src` directory under `build`.
- 
-This will:
+```bash
+gcc -O2 -Wall -fno-align-functions -DUNIXCONSOLE src/trf.c -o trf -lm
+```
+
+For more details on simplified building, see [BUILD.md](BUILD.md).
+
+### Traditional Build with Autotools ###
 
-- compile the code
-- place the executable version in the build/src directory.  
+If you prefer the traditional autotools build process:
 
-The file will be called `trf` (`trf.exe` on Windows). For backwards compatibility with automated scripts exepcting TRF to follow a certain naming scheme, the installation will also create a symbolic link named `trf<version>.<operating system>.exe`.
+```bash
+mkdir build
+cd build
+../configure
+make
+
+# To install to system
+sudo make install
+
+# Or to copy binary elsewhere
+cp src/trf DESTINATION
+```
+
+The file will be called `trf` (`trf.exe` on Windows). For backwards compatibility with automated scripts expecting TRF to follow a certain naming scheme, the installation will also create a symbolic link named `trf<version>.<operating system>.exe`.
 For example the file on a linux 64 bit operating system for version 4.10.0 will be called `trf4.10.0.linux64.exe`.
 
 ## Testing the Installation ##
-Run the executable on the included test file `test_seqs.fasta`.  (This assumes the executable has been named TRF.):
+Run the executable on the included test file `t/test_seqs.fasta`:
 
 ```bash
-trf test_seqs.fasta 2 5 7 80 10 50 2000 -l 10
+./trf t/test_seqs.fasta 2 5 7 80 10 50 2000 -l 10
 ```
 This should produce 9 files:
 ```bash
@@ -362,6 +368,12 @@ Tandem Repeats Finder finds repeats for period sizes in the range from 1 to 2000
 Some of these changes may be present in 4.09 and were undocumented, if
 they were caught during launch. They are officially present in 4.10.0.
 
+**Build improvements:**
+
+* **ARM Mac (Apple Silicon) support:** TRF now compiles and runs natively on ARM-based Macs (M1, M2, M3, etc.)
+* **Simplified build process:** New simple build script (`build.sh`) and Makefile (`Makefile.simple`) that require only gcc/clang - no autotools needed. See [BUILD.md](BUILD.md) for details.
+* **One-command compilation:** TRF can now be built with a single gcc command for easy integration into containers and build systems.
+
 Major:
 
 * TRF now has an exit status of 0 on success, and non-zero otherwise.

build.sh

@@ -0,0 +1,52 @@
+#!/bin/bash
+# Simple build script for TRF - no dependencies required
+# Just needs gcc (or clang) and standard C library
+
+set -e  # Exit on error
+
+# Detect platform
+UNAME_S=$(uname -s)
+UNAME_M=$(uname -m)
+
+case "$UNAME_S" in
+    Linux*)
+        PLATFORM="linux"
+        if [ "$UNAME_M" = "x86_64" ]; then
+            TARGET="linux64"
+        else
+            TARGET="linux32"
+        fi
+        ;;
+    Darwin*)
+        PLATFORM="mac"
+        TARGET="mac"
+        ;;
+    CYGWIN*|MINGW*)
+        PLATFORM="windows"
+        if [ "$UNAME_M" = "x86_64" ]; then
+            TARGET="dos64"
+        else
+            TARGET="dos32"
+        fi
+        ;;
+    *)
+        echo "Unknown platform: $UNAME_S"
+        exit 1
+        ;;
+esac
+
+echo "Building TRF for $PLATFORM ($TARGET)..."
+
+# Compile flags
+# Note: -fno-align-loops removed as it's not supported by clang
+CFLAGS="-O2 -Wall -fno-align-functions"
+LDFLAGS="-lm"
+
+# Use CC environment variable if set, otherwise default to gcc
+CC="${CC:-gcc}"
+
+# Build
+$CC $CFLAGS -DUNIXCONSOLE src/trf.c -o trf $LDFLAGS
+
+echo "Build complete: ./trf"
+echo "Test with: ./trf -v"

src/tr30dat.c

@@ -37,7 +37,7 @@ License along with TRF.  If not, see <https://www.gnu.org/licenses/>.
 
 
 #define new1Darrayfunc(type,functionname,length)\
-	type *functionname(int length)\
+	type *functionname(size_t length)\
 {\
 	type *objptr=calloc(length,sizeof(*objptr));\
 	if(objptr==NULL)\

src/tr30dat.h

@@ -131,7 +131,7 @@ int counterInSeq = 0;
 //#endif
 /* 02/05/16 Y. Hernandez */
 /* Since this is no longer a macro, use all lower case to avoid confusion. */
-unsigned int maxwraplength = 0;
+size_t maxwraplength = 0;
 
 /* Added by Yevgeniy Gelfand on Jan 27, 2010  */
 /* To have smaller sequences not send results */
@@ -200,7 +200,7 @@ int *Index;
 int *ACGTcount;
 
 unsigned char *Sequence;
-int            Length;
+size_t         Length;
 
 /* int S[MAXWRAPLENGTH+1][MAXPATTERNSIZE];*/
 int    Delta;   /* indel penalty */
@@ -223,10 +223,10 @@ int *Tag;    /* list of tags for linking active distances */
 int  Toptag; /* last tag in list */
 
 struct pairalign {
-    int   length;
-    int   score;
-    char *textprime, *textsecnd;
-    int * indexprime, *indexsecnd;
+    size_t length;
+    int    score;
+    char  *textprime, *textsecnd;
+    int   *indexprime, *indexsecnd;
 } AlignPair;
 
 struct cons_data {
@@ -247,8 +247,8 @@ struct cons_data {
 /*** new program started 11-29-95 **********/
 
 struct bestperiodlistelement {
-    int                           indexhigh;
-    int                           indexlow;
+    size_t                        indexhigh;
+    size_t                        indexlow;
     int                           best1;
     int                           best2;
     int                           best3;
@@ -258,14 +258,14 @@ struct bestperiodlistelement {
 } Bestperiodlist[1];
 
 struct distanceentry {
-    int location;
+    size_t location;
     int size;
 };
 
 struct distancelist {
     int k_run_sums_criteria, waiting_time_criteria, lo_d_range, hi_d_range;
     int numentries, nummatches;
-    int lowindex, highindex;
+    size_t lowindex, highindex;
     int linked;
     int linkdown, linkup;
     struct distanceentry *entry;
@@ -279,18 +279,18 @@ struct distancelist {
    alignment in the same region with the same distance */
 
 struct distanceseenarrayelement {
-    int index;
-    int end;
+    size_t index;
+    size_t end;
     int score;
 } * Distanceseenarray;
 
 struct distancelistelement {
-    int index;
+    size_t index;
     int distance;
     int changed_from_distance; /* use for test in
                                   search_for_distance_match_in_distanceseenlist
                                   3/10/05 */
-    int end;
+    size_t end;
     int score;
     int best_possible_score; /* number of copies X length X match weight */
     int accepted;
@@ -360,7 +360,7 @@ typedef struct {
     int          redundoff;
     int          ngs;
     int          use_stdin;
-    unsigned int maxwraplength;
+    size_t       maxwraplength;
 
     char  inputfilename[_MAX_PATH]; /* constant defined in stdlib */
     char  outputprefix[_MAX_PATH];
@@ -491,11 +491,11 @@ int print_flanking = 0;
 
 typedef struct {
     /* Changed to unsigned Feb 16, 2016 Yozen */
-    unsigned int length;
-    int          composition[26];
-    int          nucleotides;
-    char         name[MAXSEQNAMELEN];
-    char *       sequence;
+    size_t length;
+    int    composition[26];
+    int    nucleotides;
+    char   name[MAXSEQNAMELEN];
+    char  *sequence;
 
 } FASTASEQUENCE;
 

src/trf.c

@@ -81,6 +81,7 @@ char* GetNamePartAddress(char* name);
 void PrintBanner(void);
 static int ParseInt(const char *str, int *dest);
 static int ParseUInt(const char *str, unsigned int *dest);
+static int ParseSize(const char *str, size_t *dest);
 
 int main(int ac, char** av)
 {
@@ -204,7 +205,7 @@ int main(int ac, char** av)
 				exit(2);
 			}
 
-			if (ParseUInt(av[8], &paramset.maxwraplength) == 0) {
+			if (ParseSize(optarg, &paramset.maxwraplength) == 0) {
 				fprintf(stderr, "Error while parsing max TR length (option '-L') value\n");
 				PrintBanner();
 				exit(1);
@@ -368,6 +369,20 @@ static int ParseUInt(const char *str, unsigned int *dest)
     }
 }
 
+static int ParseSize(const char *str, size_t *dest)
+{
+    errno = 0;
+    char *temp;
+    unsigned long long val = strtoull(str, &temp, 0);
+
+    if (temp == str || *temp != '\0' || (val == ULLONG_MAX && errno == ERANGE)) {
+   		return 0;
+	}
+
+	*dest = val;
+    return 1;
+}
+
 void    PrintBanner(void)
 {
 	fprintf(stderr,"\nTandem Repeats Finder, Version %s", versionstring);

src/trfclean.h

@@ -174,7 +174,7 @@ IL*     GetList(char * datafile)
 	for (counter =0; counter<3; counter++) fgets(hparameters, 255, fp);
 
 	/* get hlength from another global variable (bad practice)*/
-	sprintf(hlength, "Length:  %d", Length);
+	sprintf(hlength, "Length:  %zu", Length);
 
 	/* loop to fill out list from buffer */
 	counter = 1; /* keeps track of order they are found */