Add files via upload

The information in the output files is improved by separating the mapping of unmethylated Cs from non-hydroxymethylated Cs. Same action with nonC
grev-uv · Jul 9, 2021 · 6caa5c8 · 6caa5c8
1 parent e0e44fc
commit 6caa5c8
Show file tree

Hide file tree

Showing 9 changed files with 40 additions and 14 deletions.
diff --git a/hmapper/src/consumer.c b/hmapper/src/consumer.c
@@ -157,21 +157,25 @@ int consumer_stage_step(void* data, scheduler_input_t* scheduler) {
 void consumer_meth_array_serialize(meth_array_node_t* array, size_t length, 
         FILE* fd, char delimiter, char record_delimiter, size_t coverage) {
   uint32_t position = 0;
-  uint16_t c = 0, nc = 0, mc = 0, hmc = 0;
+  uint16_t c = 0, nc = 0, mc = 0, ch = 0, nch = 0, hmc = 0;
 
   for (size_t i = 0; i < length; ++i) {
     position = array[i].position;
-    c = array[i].c_count;
-    nc = array[i].nc_count;
-    mc = array[i].mc_count;
+    c   = array[i].c_count;
+    nc  = array[i].nc_count;
+    mc  = array[i].mc_count;
+    ch  = array[i].ch_count;
+    nch = array[i].nch_count;
     hmc = array[i].hmc_count;
 
-    if (c + mc > coverage || c + hmc > coverage) {
-      fprintf(fd, "%u%c%u%c%u%c%u%c%u%c", 
+    if (c + mc > coverage || ch + hmc > coverage) {
+      fprintf(fd, "%u%c%u%c%u%c%u%c%u%c%u%c%u%c", 
         position, delimiter, 
         c, delimiter, 
         nc, delimiter, 
         mc, delimiter, 
+        ch, delimiter,
+        nch, delimiter,
         hmc, record_delimiter);
     }
   }
@@ -188,7 +192,7 @@ void consumer_meth_array_serialize_mix(meth_array_node_t* array_f,
                                        char record_delimiter, 
                                        size_t coverage) {
   uint32_t position = 0;
-  uint16_t c = 0, nc = 0, mc = 0, hmc = 0;
+  uint16_t c = 0, nc = 0, mc = 0, ch = 0, nch = 0, hmc = 0;
 
   uint32_t idx_f = 0, idx_r = 0;
   uint32_t position_f = 0, position_r = 0;
@@ -206,6 +210,8 @@ void consumer_meth_array_serialize_mix(meth_array_node_t* array_f,
       c        = array_f[idx_f].c_count;
       nc       = array_f[idx_f].nc_count;
       mc       = array_f[idx_f].mc_count;
+      ch       = array_f[idx_f].ch_count;
+      nch      = array_f[idx_f].nch_count;
       hmc      = array_f[idx_f].hmc_count;
 
       idx_f++;
@@ -215,6 +221,8 @@ void consumer_meth_array_serialize_mix(meth_array_node_t* array_f,
       c        = array_r[idx_r].c_count;
       nc       = array_r[idx_r].nc_count;
       mc       = array_r[idx_r].mc_count;
+      ch       = array_r[idx_f].ch_count;
+      nch      = array_r[idx_f].nch_count;
       hmc      = array_r[idx_r].hmc_count;
 
       idx_r++;
@@ -224,19 +232,23 @@ void consumer_meth_array_serialize_mix(meth_array_node_t* array_f,
       c        = array_r[idx_r].c_count   + array_f[idx_f].c_count;
       nc       = array_r[idx_r].nc_count  + array_f[idx_f].nc_count;
       mc       = array_r[idx_r].mc_count  + array_f[idx_f].mc_count;
+      ch       = array_r[idx_r].ch_count  + array_f[idx_f].ch_count;
+      nch      = array_r[idx_r].nch_count + array_f[idx_f].nch_count;
       hmc      = array_r[idx_r].hmc_count + array_f[idx_f].hmc_count;
 
       idx_f++;
       idx_r++;
     }
 
     // write the position and coverage in file
-    if (c + mc > coverage || c + hmc > coverage) {
-      fprintf(fd, "%u%c%u%c%u%c%u%c%u%c", 
+    if (c + mc > coverage || ch + hmc > coverage) {
+      fprintf(fd, "%u%c%u%c%u%c%u%c%u%c%u%c%u%c", 
         position, delimiter, 
         c, delimiter, 
         nc, delimiter, 
         mc, delimiter, 
+        ch, delimiter,
+        nch, delimiter,
         hmc, record_delimiter);
     }
 

diff --git a/hmapper/src/consumer.o b/hmapper/src/consumer.o
diff --git a/hmapper/src/main.o b/hmapper/src/main.o
diff --git a/hmapper/src/options.o b/hmapper/src/options.o
diff --git a/hmapper/src/producer.o b/hmapper/src/producer.o
diff --git a/hmapper/src/scheduler.o b/hmapper/src/scheduler.o
diff --git a/hmapper/src/worker.c b/hmapper/src/worker.c
@@ -318,16 +318,28 @@ void worker_process_alignment(worker_input_t* worker, alignment_t* alignment, si
               current_node->hmc_count++;
             }
           } else {
-            current_node->c_count++;
+            if (alignment->methylation_type == MC_QUEUE_INDEX) {
+              current_node->c_count++;
+            } else {
+              current_node->ch_count++;
+            }
           }
         } else {
           if (strand == STRAND_FORWARD) {
             if (seq_base != 'C') {
-              current_node->nc_count++;
+              if (alignment->methylation_type == MC_QUEUE_INDEX) {
+                current_node->nc_count++;
+              } else {
+                current_node->nch_count++;
+              }
             }
           } else {
             if (seq_base != 'G') {
-              current_node->nc_count++;
+              if (alignment->methylation_type == MC_QUEUE_INDEX) {
+                current_node->nc_count++;
+              } else {
+                current_node->nch_count++;
+              }
             }
           }
         }

diff --git a/hmapper/src/worker.h b/hmapper/src/worker.h
@@ -20,9 +20,11 @@
  */
 typedef struct __attribute__((__packed__)) meth_array_node {
   uint32_t position;    /* Position in the chromosome */
-  uint16_t c_count;      /* Number of total cytosines  */
-  uint16_t nc_count;     /* Number of non-methylated cytosines */
+  uint16_t c_count;      /* Number of non-methylated cytosines */
+  uint16_t nc_count;     /* Number of overlaping non-cytosines (5mC)*/
   uint16_t mc_count;     /* Number of 5-mC cytosines   */
+  uint16_t ch_count;     /* Number of non-hydroxymethylated cytosines */
+  uint16_t nch_count;    /* Number of overlaping non-cytosines (5hmC)*/
   uint16_t hmc_count;    /* Number of 5-hmC cytosines  */
 } meth_array_node_t;
 

diff --git a/hmapper/src/worker.o b/hmapper/src/worker.o