+ small changes
[qpalma.git] / tools / data_tools / filterReads.c
index 6044142..aea1b25 100644 (file)
@@ -7,37 +7,59 @@
 ////////////////////////////////////////////////////////////
 
 #include <sys/mman.h>
+#include <sys/stat.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <unistd.h>
+#include <math.h>
 
 #include "datastructures.h"
 
+#define _FILE_OFFSET_BITS == 64
+
+int compare_gene_struct(struct gene* a, struct gene* b) {
+   return a->stop - b->start;
+}
+
 int parse_gff(char* filename,FILE* fid,struct gene*** allGenes);
 
-void process_reads(FILE* reads_fs,struct gene*** allGenes,int numGenes);
+void sort_genes(struct gene*** allGenes, int numGenes);
+
+void process_reads(FILE* reads_fs,struct gene*** allGenes,int numGenes,FILE* out_fs);
+
+void combine_info(int exon_stop, int exon_start, void** upstream, int up_size, void** downstream, int down_size, FILE* out_fs,const char* gene_id);
+
+//int fitting(char* up_prb, char* up_prb_end, char* down_prb, char* down_prb_end);
+int fitting(char* up_prb, char* down_prb);
+
+void remove_ambiguities(char * old_seq, int old_seq_size, char* new_seq);
+
+static char *info = "Usage is:\n./filterReads gff reads output";
 
-static char *info = "Usage is:\n./filterReads gff reads";
+const int read_size = 36;
 
+int combined_reads = 0;
 int main(int argc, char* argv[]) {
 
-   if(argc != 3) {
+   if(argc != 4) {
       printf("%s\n",info);
       exit(EXIT_FAILURE);
    }
 
-   //static size_t page_size;
-   // page_size = (size_t) sysconf (_SC_PAGESIZE);
    int status;
    int filenameSize = 256;
-   char *gff_filename = malloc(sizeof(char)*filenameSize);
-   char *reads_filename = malloc(sizeof(char)*filenameSize);
+   char* gff_filename = malloc(sizeof(char)*filenameSize);
+   char* reads_filename = malloc(sizeof(char)*filenameSize);
+   char* output_filename = malloc(sizeof(char)*filenameSize);
 
    strncpy(gff_filename,argv[1],filenameSize);
    strncpy(reads_filename,argv[2],filenameSize);
+   strncpy(output_filename,argv[3],filenameSize);
 
    FILE *gff_fs = fopen(gff_filename,"r");
    FILE *reads_fs = fopen(reads_filename,"r");
+   FILE *out_fs = fopen(output_filename,"w");
 
    if(gff_fs == NULL) {
       printf("Error: Could not open file: %s",gff_filename);
@@ -49,36 +71,57 @@ int main(int argc, char* argv[]) {
       exit(EXIT_FAILURE);
    }
 
+   if(out_fs == NULL) {
+      printf("Error: Could not open file: %s",output_filename);
+      exit(EXIT_FAILURE);
+   }
+
    struct gene** allGenes;
    int numGenes = parse_gff(gff_filename,gff_fs,&allGenes);
    status = fclose(gff_fs);
+   free(gff_filename);
    if(status != 0)
       printf("closing of gff filestream failed!\n");
 
-   process_reads(reads_fs,&allGenes,numGenes);
+   printf("Successfully parsed gff file! Found %d genes.\n",numGenes);
 
-   status += fclose(reads_fs);
+   // some entries in the gff files are not in a sorted order
+   //printf("Sorting genes...\n");
+   //sort_genes(&allGenes,numGenes);
+   //qsort(allGenes,numGenes,sizeof(struct gene*),compare_gene_struct);
+   ///printf("Genes were sorted!\n");
+   
+   int gidx, eidx;
+   int exon_cov = 0;
+   for(gidx=0;gidx<numGenes;gidx++) {
+      for(eidx=0;eidx<allGenes[gidx]->num_exons;eidx++) {
+         exon_cov += allGenes[gidx]->exon_stops[eidx] - allGenes[gidx]->exon_starts[eidx];
+   }}
+   printf("Exon coverage is %f\n",(double)exon_cov/30432563);
+
+   process_reads(reads_fs,&allGenes,numGenes,out_fs);
+
+   status = fclose(reads_fs);
+   status = fclose(out_fs);
    if(status != 0)
-      printf("closing of filestreams failed!\n");
+      perror("fclose");
       
-   //free(allGenes);
-   free(gff_filename);
    free(reads_filename);
+   free(output_filename);
    return 0;
 }
 
-
-void process_reads(FILE* reads_fs,struct gene*** allGenes,int numGenes) {
+void process_reads(FILE* reads_fs,struct gene*** allGenes,int numGenes, FILE* out_fs) {
    int status;
 
    int buffer_size= 64;
    int chr        = 0;
    int pos        = 0;
    char* seq      = malloc(sizeof(char)*buffer_size);
-   int id         = 0;
+   unsigned long id         = 0;
    char strand    = ' ';
    int mismatch   = 0;
-   int repetition = 0;
+   int occurrence = 0;
    int size       = 0;
    int cut        = 0;
    char* prb      = malloc(sizeof(char)*buffer_size);
@@ -86,151 +129,505 @@ void process_reads(FILE* reads_fs,struct gene*** allGenes,int numGenes) {
    char* chastity = malloc(sizeof(char)*buffer_size);
 
    int reads_fid = fileno(reads_fs);
-   size_t mmapAreaSize = 20000;
+   struct stat reads_stat;
+   if ( fstat(reads_fid,&reads_stat) == -1) {
+      perror("fstat");
+      exit(EXIT_FAILURE);
+   }
+   off_t reads_filesize = reads_stat.st_size;
+   printf("Reads file is of size %lu bytes\n",(unsigned long) reads_filesize);
+   int numReads = reads_filesize / 178.0;
 
-   void *reads_area = mmap (NULL,mmapAreaSize,PROT_READ|PROT_WRITE,MAP_PRIVATE,reads_fid,0);
-   if((long int) reads_area == -1)
-      printf("mmapping failed!\n");
+   void *reads_area = mmap (NULL,reads_filesize,PROT_READ|PROT_WRITE,MAP_PRIVATE,reads_fid,0);
+   if (reads_area == MAP_FAILED) {
+      perror("mmap");
+      exit(EXIT_FAILURE);
+   }
+   close(reads_fid);
+   printf("Successfully mapped %lu bytes of reads file into memory\n",(unsigned long)reads_filesize);
 
    void* linePtr = reads_area;
-
-   char *test_string = malloc(sizeof(char)*10);
-   strncpy(test_string,(char*)reads_area,10);
-   
-   status = munmap(reads_area,mmapAreaSize);
-   if(status != 0)
-      printf("munmap failed!\n");
-
-   void** upstream_end;
-   void** upstream_overlap;
-
-   void** downstream_start;
-   void** downstream_overlap;
-   
+   char* current_line = malloc(sizeof(char)*256);
+
+   int SIZE = 500;
+   // initialize boundary arrays
+   void** upstream_end      = malloc(sizeof(void*)*SIZE);
+   void** upstream_overlap  = malloc(sizeof(void*)*SIZE);
+   void** downstream_start  = malloc(sizeof(void*)*SIZE);
+   void** downstream_overlap= malloc(sizeof(void*)*SIZE);
    int ue,uo,ds,dov;
-      
-   int SIZE = 50;
+   ue = uo = ds = dov = 0;
 
    int skippedLinesCounter = 0;
 
-   printf("Entering loop...\n");
+   int prev_exon_start = -1;
+   int prev_exon_stop = -1;
+   int cur_exon_start = -1;
+
+   current_line = strncpy(current_line,linePtr,256);
+   int gene_idx = 0;
+   int exon_idx = 1;
+   struct gene* currentGene = (*allGenes)[gene_idx];
+   char* gene_id = currentGene->id;
+
+   char* disamb_seq = malloc(sizeof(char)*read_size);
+
+   int skippedReadCtr = 0;
+   int uselessReadCtr = 0;
+   int exonicReadCtr = 0;
+   int endPrevCtr = 0;
+   int prevOverlapCtr = 0;
+   int currentStartCtr = 0;
+   int currentOverlapCtr = 0;
+   int multioccurReadCtr = 0;
+
+   int readCtr = 0;
+   // start of the parsing loop 
+   while(1) {
+      if (gene_idx == numGenes || strcmp(current_line,"") == 0)
+         break;
+
+      gene_id = currentGene->id;
+
+      if (readCtr != 0 && readCtr % 1000000 == 0)
+         printf("Processed %d/%d genes and %d/%d reads.\n",gene_idx,numGenes,readCtr,numReads);
+
+      //if (gene_idx >= 1833)
+      //   printf("currentGene start/stop: %d/%d. Positions is %d\n",currentGene->start,currentGene->stop,pos);
+
+      status = sscanf(current_line,"%d\t%d\t%s\t%lu\t%c\t%d\t%d\t%d\t%d\t%s\t%s\t%s\n",
+      &chr,&pos,seq,&id,&strand,&mismatch,&occurrence,&size,&cut,prb,cal_prb,chastity);
+      if (status < 12) {
+         skippedLinesCounter++;
+         goto next_read;
+      }
 
-   int gene_idx;
-   for (gene_idx = 0; gene_idx < numGenes;) {
-      struct gene* currentGene = (*allGenes)[gene_idx];
+      // if the read is occurring several times elsewhere then get rid of it 
+      //if(!(occurrence >= 1 && occurrence <= 25)) {
+      if ( occurrence != 1 ) {
+         multioccurReadCtr++;
+         goto next_read;
+      }
 
-      int exon_idx;
-      for(exon_idx=1;exon_idx<currentGene->num_exons-1;exon_idx++) {
-         int prev_exon_stop = currentGene->exon_stops[exon_idx-1];
-         int cur_exon_start = currentGene->exon_starts[exon_idx];
+      if (!(currentGene->start <= pos && (pos + read_size-1) <= currentGene->stop)) { // read is not within gene borders
 
-         if (cur_exon_start - prev_exon_stop < 6)
+         if ( currentGene->stop < (pos + read_size-1) ) { // go to next gene
+            gene_idx++;
+            exon_idx = 1;
+            currentGene = (*allGenes)[gene_idx];
+            //printf("currentGene->start / currentGene->stop %d/%d pos is %d\n",currentGene->start,currentGene->stop,pos);
+            ue = uo = ds = dov = 0;
             continue;
+         }
+
+         if ( pos < currentGene->start ) { // go to next read
+            skippedReadCtr++;
+            
+            next_read:
 
-         // initialize boundary arrays
-         upstream_end      = malloc(sizeof(void*)*SIZE);
-         upstream_overlap  = malloc(sizeof(void*)*SIZE);
-         downstream_start  = malloc(sizeof(void*)*SIZE);
-         downstream_overlap= malloc(sizeof(void*)*SIZE);
-         ue = 0;
-         uo = 0;
-         ds = 0;
-         dov = 0;
+            while (*(char*)linePtr != '\n') linePtr++;
+            linePtr++;
+            readCtr += 1;
+            current_line = strncpy(current_line,linePtr,256);
+            continue;
+         }
 
-         printf("Fetching reads line...\n");
+      } else { // read IS within gene borders
 
-         label:
+         exon_label:
 
-         status = sscanf((char*)linePtr,"%d\t%d\t%s\t%d\t%c\t%d\t%d\t%d\t%d\t%s\t%s\t%s\n",
-         &chr,&pos,seq,&id,&strand,&mismatch,&repetition,&size,&cut,prb,cal_prb,chastity);
-         if (status < 12) {
-            skippedLinesCounter++;
+         if (exon_idx == currentGene->num_exons) {
+            gene_idx++;
+            exon_idx = 1;
+            currentGene = (*allGenes)[gene_idx];
             continue;
          }
 
-         printf("read line!");
+         prev_exon_start = currentGene->exon_starts[exon_idx-1];
+         prev_exon_stop = currentGene->exon_stops[exon_idx-1];
+         cur_exon_start = currentGene->exon_starts[exon_idx];
 
-         if (currentGene->start <= pos && pos+35 <= currentGene->stop) {
-            printf("gene boundaries: %d %d, pos: %d\n",currentGene->start,currentGene->stop,pos);
+         //printf("exon %d %d inton til %d pos %d\n",prev_exon_start,prev_exon_stop,cur_exon_start,pos);
+            
+         if (cur_exon_start - prev_exon_stop < 6 || cur_exon_start < pos ) { // go to next exon
+            exon_idx++;
 
-            // upstream ending
-            if (pos+35 == prev_exon_stop) {
-               upstream_end[ue] = linePtr;
-               ue++;
-            }
+            if (ue != 0 && dov != 0)
+               combine_info(prev_exon_stop,cur_exon_start,upstream_end,ue,downstream_overlap,dov,out_fs,gene_id);
 
-            // upstream overlapping
-            if (prev_exon_stop - pos <= 30) {
-               upstream_overlap[uo] = linePtr;
-               uo++;
-            }
+            if (uo != 0 && ds != 0) 
+               combine_info(prev_exon_stop,cur_exon_start,upstream_overlap,uo,downstream_start,ds,out_fs,gene_id);
 
-            // downstream starting
-            if (pos == cur_exon_start) {
-               downstream_start[ds] = linePtr;
-               ds++;
-            }
+            ue = uo = ds = dov = 0;
+            goto exon_label;
+         }
 
-            // downstream overlapping
-            if (cur_exon_start - pos >= 6) {
-               downstream_overlap[dov] = linePtr;
-               dov++;
-            }
+         if ( prev_exon_start < pos && (pos+read_size) < prev_exon_stop ) { // read is inside previous exon
+            exonicReadCtr++;
 
-            while (*(char*)linePtr != '\n') {
-               printf("linePtr is %d",linePtr);
-               linePtr++;
-            }
+   // Removed exonic reads
 
-            linePtr++;
-            goto label;
+            //remove_ambiguities(seq,strlen(seq),disamb_seq);
+            //fprintf(out_fs,"%d\t%c\t%s\t%d\t%s\t%s\t%s\n",chr,strand,disamb_seq,read_size,prb,cal_prb,chastity);
 
-            free(upstream_end);
-            free(upstream_overlap);
-            free(downstream_start);
-            free(downstream_overlap);
+            goto next_read;
          }
 
-         while (*(char*)linePtr != '\n') {
-            printf("linePtr is %d",linePtr);
-            linePtr++;
+         if ( pos + (read_size-1) < prev_exon_stop ) // go to next read
+            goto next_read;
+
+         // if this position is reached the read is somehow overlapping or
+         // exactly on the exon boundary. now determine where exactly:
+         if (pos + (read_size-1) == prev_exon_stop) {  // read ends at previous exon end
+            endPrevCtr++;
+            upstream_end[ue] = linePtr;
+            ue++;
+            goto next_read;
          }
 
-         linePtr++;
-         goto label;
-         
-      }
+         if ( (prev_exon_stop - pos) >= 6 && (prev_exon_stop - pos) <= 30) { // read overlaps with previous exon boundary
+            prevOverlapCtr++;
+            upstream_overlap[uo] = linePtr;
+            uo++;
+            goto next_read;
+         }
+
+         if ( pos == cur_exon_start ) { // read starts at current exon start
+            currentStartCtr++;
+            downstream_start[ds] = linePtr;
+            ds++;
+            goto next_read;
+         }
+
+         if ( (cur_exon_start - pos) >= 6 && (cur_exon_start - pos) <= 30 ) { // read overlaps with current exon boundary
+            currentOverlapCtr++;
+            downstream_overlap[dov] = linePtr;
+            dov++;
+            goto next_read;
+         }
 
-      gene_idx++;
+         uselessReadCtr++;
+         goto next_read; // read was not useful i.e. not overlapping/starting at exon boundaries
+      }
    }
 
-   printf("skipped %d lines.\n",skippedLinesCounter);
+   combine_info(prev_exon_stop,cur_exon_start,upstream_end,ue,downstream_overlap,dov,out_fs,gene_id);
+   combine_info(prev_exon_stop,cur_exon_start,upstream_overlap,uo,downstream_start,ds,out_fs,gene_id);
+
+   free(upstream_end);
+   free(upstream_overlap);
+   free(downstream_start);
+   free(downstream_overlap);
+
+   printf("Processed %d reads in total while %d reads where skipped.\n",readCtr,skippedLinesCounter);
+   printf("%d were totally intronic. %d were outside of genes and %d were occurring on several positions.\n",uselessReadCtr,skippedReadCtr,multioccurReadCtr);
+   printf("%d reads were useless in total\n",uselessReadCtr+skippedReadCtr+multioccurReadCtr);
+   printf("%d reads were totally exonic\n",exonicReadCtr);
+   printf("%d reads overlap with prev. exon. %d reads overlap with current exon\n",prevOverlapCtr,currentOverlapCtr);
+   printf("%d reads where newly combined from two original reads\n",combined_reads);
+   printf("Total used reads: %d\n",exonicReadCtr+endPrevCtr+prevOverlapCtr+currentStartCtr+currentOverlapCtr);
 
+   status = munmap(reads_area,reads_filesize);
+   if(status != 0)
+      perror("munmap");
+
+   //free(current_line);
    free(seq);
    free(prb);
    free(cal_prb);
    free(chastity);
 }
 
-//      int skippedLinesCounter = 0;
-//      while(1) {
-//         status = fscanf(reads_fs,"%d\t%d\t%s\t%d\t%c\t%d\t%d\t%d\t%d\t%s\t%s\t%s\n",&chr,&pos,seq,&id,&strand,&mismatch,&repetition,&size,&cut,prb,cal_prb,chastity);
-//         if(status == EOF)
-//            break;
-//
-//         if (status < 12) {
-//            skippedLinesCounter++;
-//            continue;
-//         }
-//
-//         label:
-//
-//         if (pos < currentGene->start)
-//            continue;
-//
-//         if (currentGene->stop < pos) {
-//            gene_idx++;
-//            currentGene = (*allGenes)[gene_idx];
-//            goto label;
-//         }
-//
+void combine_info(int exon_stop, int exon_start, void** upstream, int up_size, void** downstream, int down_size,FILE* out_fs,const char* gene_id) {
+   //printf("up/down size is %d/%d\n",up_size,down_size);
+   
+   int up_idx, down_idx, status;
+   char* upstream_line = malloc(sizeof(char)*256);
+   char* downstream_line = malloc(sizeof(char)*256);
+
+   int buffer_size= 64;
+
+   int up_chr        = 0;
+   int up_pos        = 0;
+   char* up_seq      = malloc(sizeof(char)*buffer_size);
+   int up_id         = 0;
+   char up_strand    = ' ';
+   int up_mismatch   = 0;
+   int up_occurrence = 0;
+   int up_sz       = 0;
+   int up_cut        = 0;
+   char* up_prb      = malloc(sizeof(char)*buffer_size);
+   char* up_cal_prb  = malloc(sizeof(char)*buffer_size);
+   char* up_chastity = malloc(sizeof(char)*buffer_size);
+
+   int down_chr        = 0;
+   int down_pos        = 0;
+   char* down_seq      = malloc(sizeof(char)*buffer_size);
+   int down_id         = 0;
+   char down_strand    = ' ';
+   int down_mismatch   = 0;
+   int down_occurrence = 0;
+   int down_sz         = 0;
+   int down_cut        = 0;
+   char* down_prb      = malloc(sizeof(char)*buffer_size);
+   char* down_cal_prb  = malloc(sizeof(char)*buffer_size);
+   char* down_chastity = malloc(sizeof(char)*buffer_size);
+
+   int new_chr        = 0;
+   char* new_seq      = malloc(sizeof(char)*buffer_size);
+   char new_strand    = ' ';
+   char* new_prb        = malloc(sizeof(char)*buffer_size);
+   char* new_cal_prb    = malloc(sizeof(char)*buffer_size);
+   char* new_chastity   = malloc(sizeof(char)*buffer_size);
+   char* new_up_seq = malloc(sizeof(char)*read_size);
+   char* new_down_seq = malloc(sizeof(char)*read_size); 
+
+   int overlap;
+   int fit;
+
+   char* used_flag = calloc(down_size,sizeof(char));
+
+   for(up_idx=0;up_idx<up_size;up_idx++) {
+      strncpy(upstream_line,upstream[up_idx],256);
+      status = sscanf(upstream_line,"%d\t%d\t%s\t%d\t%c\t%d\t%d\t%d\t%d\t%s\t%s\t%s\n",
+      &up_chr,&up_pos,up_seq,&up_id,&up_strand,&up_mismatch,&up_occurrence,&up_sz,
+      &up_cut,up_prb,up_cal_prb,up_chastity);
+
+      remove_ambiguities(up_seq,strlen(up_seq),new_up_seq);
+
+      overlap = exon_stop - up_pos;
+
+      for(down_idx=0;down_idx<down_size;down_idx++) {
+         if( used_flag[down_idx] == 1)
+            continue;
+
+         strncpy(downstream_line,downstream[down_idx],256);
+         status = sscanf(downstream_line,"%d\t%d\t%s\t%d\t%c\t%d\t%d\t%d\t%d\t%s\t%s\t%s\n",
+         &down_chr,&down_pos,down_seq,&down_id,&down_strand,&down_mismatch,&down_occurrence,&down_sz,
+         &down_cut,down_prb,down_cal_prb,down_chastity);
+
+         remove_ambiguities(down_seq,strlen(down_seq),new_down_seq);
+
+         new_seq[0] = '\0';
+         new_prb[0] = '\0';
+         new_cal_prb[0] = '\0';
+         new_chastity[0] = '\0';
+         int splitpos = 0;
+         
+         fit = fitting(up_prb+(36-overlap),down_prb);
+         if (fit == -1)
+            continue;
+
+         if (! (fit  < overlap ))
+            continue;
+
+         new_chr     = up_chr;
+         new_strand  = up_strand;
+
+         strncat(new_seq,new_up_seq,overlap);
+         strncat(new_prb,up_prb,overlap);
+         strncat(new_cal_prb,up_cal_prb,overlap);
+         strncat(new_chastity,up_chastity,overlap);
+
+         strncat(new_seq,new_down_seq+fit,read_size-overlap);
+         strncat(new_prb,down_prb+fit,read_size-overlap);
+         strncat(new_cal_prb,down_cal_prb+fit,read_size-overlap);
+         strncat(new_chastity,down_chastity+fit,read_size-overlap);
+
+         fprintf(out_fs,"%d\t%c\t%s\t%d\t%d\t%s\t%s\t%s\t%s\n",
+         new_chr,new_strand,new_seq,splitpos,read_size,new_prb,new_cal_prb,new_chastity,gene_id);
+
+         combined_reads++;
+         used_flag[down_idx] = 1;
+      }
+   }
+
+   free(upstream_line);
+   free(downstream_line);
+  
+   free(new_up_seq);
+   free(new_down_seq); 
+
+   free(up_seq);
+   free(up_prb);
+   free(up_cal_prb);
+   free(up_chastity);
+
+   free(down_seq);
+   free(down_prb);
+   free(down_cal_prb);
+   free(down_chastity);
+
+   free(new_seq);
+   free(new_prb);
+   free(new_cal_prb);
+   free(new_chastity);
+}
+
+
+
+
+/*
+int fitting(char* up_prb, char* up_prb_end, char* down_prb, char* down_prb_end) {
+   double epsilon_mean = 15.0;
+   double epsilon_var = 10.0;
+   double mean_up = 0;
+   double variance_up = 0;
+   double mean_down = 0;
+   double variance_down = 0;
+
+   char *up_ptr = up_prb;
+   char *down_ptr = down_prb;
+
+   int w_size = 0;
+   while(up_ptr != up_prb_end) {
+      mean_up += (*up_ptr)-50;
+      mean_down += (*down_ptr)-50;
+      w_size++;
+      up_ptr++;
+      down_ptr++;
+   }
+   mean_up /= w_size;
+   mean_down /= w_size;
+
+
+   up_ptr = up_prb;
+   down_ptr = down_prb;
+   w_size = 0;
+   while(up_ptr != up_prb_end) {
+      variance_up += pow((*up_prb)-50 - mean_up,2);
+      variance_down += pow((*down_prb)-50 - mean_down,2);
+      w_size++;
+      up_ptr++;
+      down_ptr++;
+   }
+   variance_up /= (w_size-1);
+   variance_down /= (w_size-1);
+
+   //printf("means: %f %f, variances: %f %f\n",mean_up,mean_down,variance_up,variance_down);
+
+   if ( abs(mean_up - mean_down) < epsilon_mean && abs(variance_up - variance_down) < epsilon_var )
+      return 1;
+      
+   return 0;
+}
+*/
+
+int fitting(char* up_prb, char* down_prb) {
+   double epsilon_mean = 30.0;
+   double epsilon_var = 30.0;
+   int w_size = 6;
+
+   double current_mean_up = 0;
+   double current_variance_up = 0;
+   double current_mean_down = 0;
+   double current_variance_down = 0;
+
+   double* mean_up       = malloc(sizeof(double)*read_size-2*w_size);
+   double* variance_up   = malloc(sizeof(double)*read_size-2*w_size);
+   double* mean_down     = malloc(sizeof(double)*read_size-2*w_size);
+   double* variance_down = malloc(sizeof(double)*read_size-2*w_size);
+
+   int iidx = -1;
+   int uidx;
+   int didx;
+
+   for(uidx=iidx-w_size;uidx<iidx;uidx++) {
+      didx = uidx+w_size;
+      current_mean_up += up_prb[uidx]-50;
+      current_mean_down += up_prb[didx]-50;
+
+   }
+   current_mean_up /= w_size;
+   current_mean_down /= w_size;
+
+   for(uidx=iidx-w_size;uidx<iidx;uidx++) {
+      didx = uidx+w_size;
+      current_variance_up += pow(up_prb[uidx]-50 - current_mean_up,2);
+      current_variance_down += pow(up_prb[didx]-50 - current_mean_up,2);
+      
+   }
+   current_variance_up /= (w_size-1);
+   current_variance_down /= (w_size-1);
+
+   for(iidx=w_size;iidx<30;iidx++) {
+      for(uidx=iidx-w_size;uidx<iidx;uidx++) {
+         didx = uidx+w_size;
+         mean_up[iidx-w_size] += down_prb[uidx]-50;
+         mean_down[iidx-w_size] += down_prb[didx]-50;
+
+      }
+      mean_up[iidx-w_size] /= w_size;
+      mean_down[iidx-w_size] /= w_size;
+
+      for(uidx=iidx-w_size;uidx<iidx;uidx++) {
+         didx = uidx+w_size;
+         variance_up[iidx-w_size] += pow(down_prb[uidx]-50 - mean_up[iidx-w_size],2);
+         variance_down[iidx-w_size] += pow(down_prb[didx]-50 - mean_down[iidx-w_size],2);
+      }
+      variance_up[iidx-w_size] /= (w_size-1);
+      variance_down[iidx-w_size] /= (w_size-1);
+
+      //printf("means: %f %f %f %f, variances: %f %f %f %f\n",current_mean_up,current_mean_down,mean_up,mean_down,current_variance_up,current_variance_down,variance_up,variance_down);
+      //if ( abs(current_mean_up - mean_up) < epsilon_mean && abs(current_variance_up - variance_up) < epsilon_var 
+      //&& abs(current_mean_down - mean_down) < epsilon_mean && abs(current_variance_down - variance_down) < epsilon_var )
+      //   return iidx;
+   }
+
+   int bidx;
+   int bestIdx = -1;
+   double min = 1000.0;
+   for(bidx=0;bidx<read_size-2*w_size;bidx++) {
+      if ( abs(current_mean_up - mean_up[bidx]) < epsilon_mean && abs(current_variance_up - variance_up[bidx]) < epsilon_var 
+      && abs(current_mean_down - mean_down[bidx]) < epsilon_mean && abs(current_variance_down - variance_down[bidx]) < epsilon_var ) {
+         if ( abs(current_mean_up - mean_up[bidx]) + abs(current_variance_up - variance_up[bidx]) + abs(current_mean_down - mean_down[bidx]) + abs(current_variance_down - variance_down[bidx])) {
+            min = abs(current_mean_up - mean_up[bidx]) + abs(current_variance_up - variance_up[bidx]) + abs(current_mean_down - mean_down[bidx]) + abs(current_variance_down - variance_down[bidx]);
+            bestIdx = bidx;
+         }
+      }
+   }
+
+   free(mean_up);
+   free(variance_up);
+   free(mean_down);
+   free(variance_down);
+
+   return bestIdx;
+}
+
+void remove_ambiguities(char * old_seq, int old_seq_size, char* new_seq) {
+   int idx=0;
+   int new_idx = 0;
+
+   while(idx<old_seq_size) {
+      //printf("Current elem %c pos %d %d\n",old_seq[idx],idx,new_idx);
+      if (old_seq[idx] == ']' || old_seq[idx] == '-' ) {
+         idx++;
+         continue;
+      }
+
+      if (old_seq[idx] == '[') {
+         idx += 2;
+         //printf("%c %c\n",old_seq[idx-2],old_seq[idx]);
+         continue;
+      }
+
+      new_seq[new_idx] = old_seq[idx];
+      idx++;
+      new_idx++;
+   }
+
+   if (new_idx != 36) {
+      printf("Error: Sequence is not of length 36!\n");
+      printf("old seq: %s\n",old_seq);
+      printf("new seq: %s\n",new_seq);
+      exit(EXIT_FAILURE);
+   }
+}
+
+/*
+ * TODO:
+ * - Check strand -> done simple (only if equal)
+ * - check for [AC] and similar entries -> done simple (see function
+ * - remove_ambiguities (exchanges [XY] by the second entry)
+ */