debian/changelog

+segemehl (0.3.4-1) unstable; urgency=medium
+
+  * New upstream version
+  * debhelper 12
+  * Standards-Version: 4.3.0
+
+ -- Andreas Tille <tille@debian.org>  Fri, 11 Jan 2019 22:52:52 +0100
+
 segemehl (0.3.2-1) unstable; urgency=medium
 
   * Fix watch file

debian/compat

-11
+12

debian/control

@@ -3,12 +3,12 @@ Maintainer: Debian Med Packaging Team <debian-med-packaging@lists.alioth.debian.
 Uploaders: Andreas Tille <tille@debian.org>
 Section: science
 Priority: optional
-Build-Depends: debhelper (>= 11),
+Build-Depends: debhelper (>= 12~),
                pkg-config,
                libhts-dev,
                libncurses-dev,
                zlib1g-dev
-Standards-Version: 4.2.1
+Standards-Version: 4.3.0
 Vcs-Browser: https://salsa.debian.org/med-team/segemehl
 Vcs-Git: https://salsa.debian.org/med-team/segemehl.git
 Homepage: http://www.bioinf.uni-leipzig.de/Software/segemehl/

include/fileio.h

@@ -45,7 +45,7 @@
 #include "stringutils.h"
 char * bl_getTempFile(char *path, char *tmp);
 int bl_UnixSort(void *space, char *filename, const char *fieldstring, const char delim);
-char* readfile(void *, char *, Uint*);
+char* readfile(void *, char *, size_t*);
 stringset_t **readcsv(void *, char *, char*, Uint *);
 void writeY(char *, double  *, Uint, Uint, Uint);
 void writeXYUint(char *filename, Uint *X, Uint *Y, Uint len);

include/segemehl.h

@@ -230,7 +230,7 @@ typedef struct segemehl_s {
     */
     merge_t *merge;
     char orientation;
- 
+    char fullname;
 } segemehl_t;
 
 typedef struct checkthread_s {
@@ -387,6 +387,7 @@ se_setdefault(segemehl_t *info) {
   info->bamhdr = NULL;
   info->head = NULL;
   info->orientation = 1;
+  info->fullname = 0;
 }
 
 inline static void

include/version.h

@@ -24,11 +24,11 @@
 
 
 
-#define VERSION "0.3.2"
+#define VERSION "0.3.4"
 
-#define REVISION "g664b3dd"
+#define REVISION "ge5dee47"
 
-#define TIME "2018-11-11 17:18:49 +0100"
+#define TIME "2018-12-25 20:39:07 +0100"
 
 
 

include/version.h.bak

@@ -12,6 +12,6 @@
 
 
 #define VERSION "v0.3.0"
-#define REVISION "g664b3dd"
-#define TIME "2018-11-11 17:18:49 +0100"
+#define REVISION "ge5dee47"
+#define TIME "2018-12-25 20:39:07 +0100"
 

libs/alignment.c

@@ -1252,7 +1252,7 @@ bl_mdGetDiffString(char *MD) {
 
 char*
 cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned char rev, char brief) {
-  Uint i, j, k, q=0, p=0, cur=0, strsize, steps, msteps;
+  Uint i, j, k, q=0, p=0, cur=0, strsize, steps=0, msteps;
   char *cigarstr;
   char eopc=0;
 
@@ -1266,10 +1266,9 @@ cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned
     cur+=strsize;
   }
 
-  for(k=0; k < al->numofmeops; k++) {
+  for(steps=0, k=0; k < al->numofmeops; k++) {
     i = (rev) ? al->numofmeops - k - 1 : k;
     //if Replacement occured 
-    steps=0;
     if (al->meops[i].eop == Replacement) {
       //iter over all steps
       msteps=0;
@@ -1299,7 +1298,7 @@ cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned
         msteps++;
       }
 
-      steps = msteps;
+      steps += msteps;
       //set string ptrs
       p+=j;
       q+=j;
@@ -1311,7 +1310,8 @@ cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned
       else
         eopc = '=';
       //set ptrs
-      steps = al->meops[i].steps;
+      //steps = al->meops[i].steps;
+      steps += al->meops[i].steps;
       p+=steps;
       q+=steps;
     }
@@ -1322,11 +1322,22 @@ cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned
       else
         eopc = 'X';
       //set ptrs
-      steps = al->meops[i].steps;
+      //steps = al->meops[i].steps;
+      steps += al->meops[i].steps;
       p+=steps;
       q+=steps;
     }
 
+    //dump previous matches and mismatches if in brief mode
+    //if a different operation is coming
+    if(brief && steps && eopc == 'M' && 
+        al->meops[i].eop != Match && al->meops[i].eop != Mismatch) {
+      
+      strsize = snprintf(NULL, 0, "%d", steps)+1;
+      sprintf(&cigarstr[cur], "%d%c", steps, eopc);
+      cur+=strsize;
+      steps = 0;
+    }
 
     //if deletion occured
     if (al->meops[i].eop == Deletion) {
@@ -1355,11 +1366,15 @@ cigarstring(Alignment *al, Uint leftclip, Uint rightclip, char clipch, unsigned
       //     p+=steps; no moves!
     }
 
-
+    //dump everything except "M"s in the brief mode
+    //unless we are at the end
+    if(eopc != 'M' || k+1 == al->numofmeops) {
       //strsize = floor(log(steps)/log(10))+2;
       strsize = snprintf(NULL, 0, "%d", steps)+1;
       sprintf(&cigarstr[cur], "%d%c", steps, eopc);
       cur+=strsize;
+      steps = 0;
+    }
   }
 
   if(rightclip || (leftclip && rev)) {

libs/fileio.c

@@ -524,14 +524,14 @@ bl_freplacestr(char *filename, char *str, Uint len, char stop){
 int
 bl_fgets(void *space, FILE *fp, char **str) {
   char ch, *buffer;
-  Uint buffersize = 100;
-  Uint len = 0;
+  size_t buffersize = MAXBUFFERSIZE;
+  size_t len = 0;
 
   buffer = ALLOCMEMORY(space, NULL, char, buffersize);
 
   while((ch=getc(fp)) != EOF && ch != '\n') {
     if(len == buffersize - 1) {
-      buffersize = 2 * buffersize + 1;
+      buffersize += MAXBUFFERSIZE + 1;
       buffer = ALLOCMEMORY(space, buffer, char, buffersize);
     }
     buffer[len++] = (char) ch;
@@ -547,13 +547,13 @@ bl_fgets(void *space, FILE *fp, char **str) {
 
 
 char* 
-readfile(void* space, char* filename, Uint* strlen) {
+readfile(void* space, char* filename, size_t* strlen) {
 
   char ch;
   char *buffer;
   FILE *fp;
-  Uint buffersize = MAXBUFFERSIZE;
-  Uint len=0;
+  size_t buffersize = MAXBUFFERSIZE;
+  size_t len=0;
 
   fp = fopen(filename, "r");
   if (fp == NULL){
@@ -565,7 +565,7 @@ readfile(void* space, char* filename, Uint* strlen) {
 
   while((ch=getc(fp)) != EOF) {
     if(len == buffersize-1) {
-      buffersize = 2*buffersize+1;
+      buffersize += MAXBUFFERSIZE+1;
       buffer = ALLOCMEMORY(space, buffer, char, buffersize);
     }
     len++;
@@ -586,7 +586,7 @@ readcsv(void *space,
     char *delim, 
     Uint *linecount) {
 
-  Uint i, contentlen;
+  size_t i, contentlen;
   char *content;
   stringset_t *lines, **csv;
 

libs/manout.c

@@ -347,8 +347,8 @@ se_output(mappingset_t *s, fasta_t * reads, unsigned int k,
         char *nextrname = samlist->recs[0].rname;
         //should always be > 0, because the alignment exists uniqly
         uint64_t nextrpos = samlist->recs[0].pos;
-        char nextrc = 0;
-        char unmappedmate = 1;
+        char nextrc = samlist->recs[0].flag & 0x10;
+        char unmappedmate = !bl_hasMateMapping(s);
         /*
          * output to device
          *
@@ -370,8 +370,8 @@ se_output(mappingset_t *s, fasta_t * reads, unsigned int k,
         char *nextrname = samlist->recs[0].rname;
         //should always be > 0, because the alignment exists uniqly
         uint64_t nextrpos = samlist->recs[0].pos;
-        char nextrc = 0;
-        char unmappedmate = 1;
+        char nextrc = samlist->recs[0].flag & 0x10; 
+        char unmappedmate = !bl_hasQueryMapping(s);
         /*
          * output to device
          *
@@ -399,11 +399,12 @@ se_output(mappingset_t *s, fasta_t * reads, unsigned int k,
       char *nextrname = NULL;
       int64_t nextrpos = 0;
       char nextrc = 0;
+      char unmappedmate = !bl_hasMateMapping(s);
       /*
        * output to device
        *
        */
-      bl_samprintEmptyAlign (desc, seq, qual, hasPaired, 1, 1, nextrname, 
+      bl_samprintEmptyAlign (desc, seq, qual, hasPaired, 1, unmappedmate, nextrname, 
           nextrpos, nextrc, 0, 0, seed, nfo);
     }
 
@@ -421,11 +422,12 @@ se_output(mappingset_t *s, fasta_t * reads, unsigned int k,
       char *nextrname = NULL;
       int64_t nextrpos = 0;
       char nextrc = 0;
+      char unmappedmate = !bl_hasQueryMapping(s);
       /*
        * output to device
        *
        */
-      bl_samprintEmptyAlign (desc, seq, qual, hasPaired, 0, 1, nextrname, 
+      bl_samprintEmptyAlign (desc, seq, qual, hasPaired, 0, unmappedmate, nextrname, 
           nextrpos, nextrc, 0, 0, mateseed, nfo);
     }
   }

libs/samio.c

@@ -57,7 +57,10 @@
  *   
  */
 
-samrec_t* bl_samInitRec(samrec_t *samrec, char *qname) {
+samrec_t* bl_samInitRec(samrec_t *samrec, char *qname, char fullname) {
+
+  char *tmp;
+  Uint len=0;
 
   samrec->qname = NULL;
   samrec->flag = 0;
@@ -73,7 +76,15 @@ samrec_t* bl_samInitRec(samrec_t *samrec, char *qname) {
   samrec->nooftags = 0;
   samrec->tags = NULL;
 
-  samrec->qname = bl_strdup(qname);
+  if(fullname) {
+    tmp = bl_strdup(qname);
+  } else {
+    len = strlen(qname);
+    tmp = strclip(NULL, qname, &len);
+  }
+
+  samrec->qname = tmp; 
+    
   return samrec;
 }
 
@@ -1339,7 +1350,7 @@ bl_samprintEmptyAlign (char *desc, char* seq, char* qual, char hasPaired,
   }
 
 
-  bl_samInitRec(&samlist->recs[0], desc);
+  bl_samInitRec(&samlist->recs[0], desc, nfo->fullname);
 
   if(!hasPaired) {
     nomatemapped = 0;
@@ -1426,7 +1437,7 @@ bl_samline2rec(char *line, Uint len, samheader_t *head) {
     switch(i+1) {
       case 1:
         qname = cur;
-        rec = bl_samInitRec(rec, qname);
+        rec = bl_samInitRec(rec, qname, 1);
         break;
       case 2: 
         flag = atoi(cur);
@@ -1640,7 +1651,7 @@ samlist_t *bl_samgetSamList (fasta_t *reads, Uint id, mapping_t *l,
       bl_reconvertBisulfite(seq, strlen(seq), nfo->bisulfite);
     }
 
-    bl_samInitRec(&samlist->recs[i], bl_getMapFragQryDesc(&l->f[i]));
+    bl_samInitRec(&samlist->recs[i], bl_getMapFragQryDesc(&l->f[i]), nfo->fullname);
 
     //predict strand
     if(nfo->split) {

libs/segemehl.c

@@ -595,6 +595,10 @@ int main(int argc, char** argv) {
       "show alignments", NULL, NULL, &info.align);
   manopt(&optset, FLAG, 0, 0, "nohead",
       "do not output header", NULL, NULL, &info.nohead);
+
+  manopt(&optset, FLAG, 0, 'f', "fullname",
+      "write full query name (no trunctation at whitespace)", NULL, NULL, &info.fullname);
+
   //manopt(&optset, FLAG, 0, 'z', "nosuflinks",
   //    "dont use suflinks (does not affect index construction, for short reads only, increases runtime!)", NULL, NULL, &info.nosuflinks);
   /*