Release/1.6.0

ToniWestbrook · Apr 1, 2024 · 17f2ca3 · 17f2ca3
1 parent 5fbfea3
commit 17f2ca3
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 9 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,9 @@
 # Change Log
 
+## [1.6.0] - 2024-03-31
+### Added
+- Alignment command can now skip ORF detection and treat input sequences as transcripts (-q option)
+
 ## [1.5.0] - 2022-11-18
 ### Changed
 - Updated PALADIN to use UniProt's new API system

diff --git a/align.c b/align.c
@@ -118,7 +118,7 @@ int command_align(int argc, char *argv[]) {
 	memset(&opt0, 0, sizeof(mem_opt_t));
     proxyAddress = NULL;
 
-	while ((c = getopt(argc, argv, "1epabgnMCSVYJjf:F:z:u:k:o:c:v:s:r:t:R:A:B:O:E:U:w:L:d:T:Q:D:m:I:N:W:x:G:h:y:K:X:H:P:")) >= 0) {
+	while ((c = getopt(argc, argv, "1epqabgnMCSVYJjf:F:z:u:k:o:c:v:s:r:t:R:A:B:O:E:U:w:L:d:T:Q:D:m:I:N:W:x:G:h:y:K:X:H:P:")) >= 0) {
 		if (c == 'k') opt->min_seed_len = atoi(optarg), opt0.min_seed_len = 1;
 		else if (c == 'u') opt->outputType = atoi(optarg);
 		else if (c == 'f') opt->min_orf_len = atoi(optarg);
@@ -147,6 +147,7 @@ int command_align(int argc, char *argv[]) {
 		else if (c == 'n') opt->proteinFlag |= ALIGN_FLAG_KEEP_PRO;
 		else if (c == 'J') opt->proteinFlag &= ~ALIGN_FLAG_ADJUST_ORF;
         else if (c == 'p') opt->proteinFlag |= ALIGN_FLAG_MANUAL_PRO;
+        else if (c == 'q') opt->proteinFlag |= ALIGN_FLAG_MANUAL_TRANSCRIPT;
 		else if (c == 'c') opt->max_occ = atoi(optarg), opt0.max_occ = 1;
 		else if (c == 'd') opt->zdrop = atoi(optarg), opt0.zdrop = 1;
 		else if (c == 'v') bwa_verbose = atoi(optarg);
@@ -278,6 +279,12 @@ int command_align(int argc, char *argv[]) {
 		}
 	} else update_a(opt, &opt0);
 
+    // Check for incompatible options
+    if ((opt->proteinFlag & ALIGN_FLAG_MANUAL_PRO) && (opt->proteinFlag & ALIGN_FLAG_MANUAL_TRANSCRIPT)) {
+		logMessage(__func__, LOG_LEVEL_ERROR, "Both protein and transcriptome mode selected.\n");
+        return 1;
+    }
+
     // Default to standard genetic code for translations
     if (!opt->translations) opt->translations = convertTransArgs("");
 
@@ -440,14 +447,14 @@ int renderAlignUsage(const mem_opt_t * passOptions) {
 	fprintf(stderr, "Usage: paladin align [options] <idxbase> <in.fq>\n\n");
 
 	fprintf(stderr, "Gene detection options:\n\n");
+    fprintf(stderr, "       -q            disable ORF detection and treat input as transcript sequence\n");
     fprintf(stderr, "       -p            disable ORF detection and treat input as protein sequence\n");
-	fprintf(stderr, "       -b            disable brute force ORF detection\n");
+	fprintf(stderr, "       -b            disable brute force detection\n");
 	fprintf(stderr, "       -J            do not adjust minimum ORF length (constant value) for shorter read lengths\n");
 	fprintf(stderr, "       -f INT        minimum ORF length accepted (as constant value) [%d]\n", passOptions->min_orf_len);
 	fprintf(stderr, "       -F FLOAT      minimum ORF length accepted (as percentage of read length) [%.2f]\n", passOptions->min_orf_percent);
     fprintf(stderr, "       -z INT[,...]  Genetic code used for translation (-z ? for full list) [1]\n");    
 
-
 	fprintf(stderr, "\nAlignment options:\n\n");
 	fprintf(stderr, "       -t INT        number of threads [%d]\n", passOptions->n_threads);
 	fprintf(stderr, "       -k INT        minimum seed length [%d]\n", passOptions->min_seed_len);

diff --git a/main.h b/main.h
@@ -65,7 +65,7 @@
 #ifndef PACKAGE_VERSION
 #define PACKAGE_VERSION STR(PACKAGE_VERSION_MAJOR) "." STR(PACKAGE_VERSION_MINOR) "." STR(PACKAGE_VERSION_REV)
 #define PACKAGE_VERSION_MAJOR 1
-#define PACKAGE_VERSION_MINOR 5
+#define PACKAGE_VERSION_MINOR 6
 #define PACKAGE_VERSION_REV 0
 #endif
 

diff --git a/protein.c b/protein.c
@@ -243,6 +243,19 @@ int getSequenceORF(char * passSequence, unsigned long passLength, int passTrans,
 		endSeqPos = getLastAlignedPos(passLength, relFrame);
 		endOrfPos = getLastAlignedOrfPos(passLength, relFrame, passOptions);
 
+        // Treat entire sequence as ORF in transcript mode
+        if (passOptions->proteinFlag & ALIGN_FLAG_MANUAL_TRANSCRIPT) {
+            addORFHistory(history, historySize, frameIdx);
+            history[0][frameIdx][historySize[frameIdx] - 1] = (strandDir == 1) ? relFrame : passLength - 1 - relFrame;
+            history[1][frameIdx][historySize[frameIdx] - 1] = getLastAlignedPos(passLength, frameIdx);
+
+            if (passOptions->proteinFlag & ALIGN_FLAG_BRUTE_ORF) {
+                continue;
+            }
+
+            break;
+        }
+
 		// Adjust min ORF length if requested
 		if ((passOptions->proteinFlag & ALIGN_FLAG_ADJUST_ORF) && (passLength < passOptions->min_orf_len)) {
 			endOrfPos = getLastAlignedPos(passLength, relFrame);

diff --git a/protein.h b/protein.h
@@ -9,11 +9,12 @@
 #define OUTPUT_TYPE_UNIPROT_SIMPLE 0
 #define OUTPUT_TYPE_UNIPROT_FULL   1
 
-#define ALIGN_FLAG_BRUTE_ORF  0x0001
-#define ALIGN_FLAG_GEN_NT     0x0002
-#define ALIGN_FLAG_KEEP_PRO   0x0004
-#define ALIGN_FLAG_ADJUST_ORF 0x0008
-#define ALIGN_FLAG_MANUAL_PRO 0x0010
+#define ALIGN_FLAG_BRUTE_ORF         0x0001
+#define ALIGN_FLAG_GEN_NT            0x0002
+#define ALIGN_FLAG_KEEP_PRO          0x0004
+#define ALIGN_FLAG_ADJUST_ORF        0x0008
+#define ALIGN_FLAG_MANUAL_PRO        0x0010
+#define ALIGN_FLAG_MANUAL_TRANSCRIPT 0x0020
 
 typedef struct {
 	unsigned long startIdx;