Accelerate overlap merge

Assembly.cpp

@@ -415,7 +415,7 @@ void Build_hash_table_multiple_thr()
 }
 
 
-void* Overlap_calculate(void* arg)
+void* Overlap_calculate_version1(void* arg)
 {
 
     int thr_ID = *((int*)arg);
@@ -517,6 +517,169 @@ void* Overlap_calculate(void* arg)
 }
 
 
+
+
+void debug_merge_result(Candidates_list* x, Candidates_list* y)
+{
+    uint64_t i;
+    if (x->length != y->length)
+    {
+        fprintf(stderr, "ERROR: different list\n");
+        fprintf(stderr, "x->length: %llu, y->length: %llu\n", x->length, y->length);
+    }
+    
+    for (i = 0; i < x->length; i++)
+    {
+        if (x->list[i].offset != y->list[i].offset 
+            ||
+            x->list[i].readID != y->list[i].readID
+            ||
+            x->list[i].self_offset != y->list[i].self_offset
+            ||
+            x->list[i].strand != y->list[i].strand
+        )
+        {
+            fprintf(stderr, "ERROR: different pos\n");
+        }
+        
+    }
+    
+}
+
+
+
+void* Overlap_calculate(void* arg)
+{
+
+    int thr_ID = *((int*)arg);
+
+    long long i = 0;
+    int avalible_k = 0;
+
+    UC_Read g_read;
+    init_UC_Read(&g_read);
+    HPC_seq HPC_read;
+    Hash_code k_code;
+    uint64_t code;
+    uint64_t end_pos;
+    k_mer_pos* list;
+    uint64_t list_length;
+    uint64_t sub_ID;
+
+    Candidates_list l;
+    //Candidates_list debug_l;
+
+    init_Candidates_list(&l);
+    //init_Candidates_list(&debug_l);
+
+    k_mer_pos_list_alloc array_list;
+    init_k_mer_pos_list_alloc(&array_list);
+
+    for (i = thr_ID; i < R_INF.total_reads; i = i + thread_num)
+    ///for (i = thr_ID; i < R_INF.total_reads/50; i = i + thread_num)
+    {
+        /**
+        if (i % 1000 == 0)
+        {
+            fprintf(stderr, "i: %llu\n", i);
+        }
+        **/
+        
+        
+
+        clear_Candidates_list(&l);
+        ///clear_Candidates_list(&debug_l);
+
+        clear_k_mer_pos_list_alloc(&array_list);
+        recover_UC_Read(&g_read, &R_INF, i);
+
+
+        ///forward strand
+        init_HPC_seq(&HPC_read, g_read.seq, g_read.length);
+        init_Hash_code(&k_code);
+        avalible_k = 0;
+
+        while ((code = get_HPC_code(&HPC_read, &end_pos)) != 6)
+        {
+            if(code < 4)
+            {
+                k_mer_append(&k_code,code,k_mer_length);
+                avalible_k++;
+                if (avalible_k>=k_mer_length)
+                {
+                    list_length = locate_Total_Pos_Table(&PCB, &k_code, &list, k_mer_length, &sub_ID);
+
+                    if (list_length != 0)
+                    {
+                        append_k_mer_pos_list_alloc(&array_list, list, list_length, end_pos, 0);
+                    }
+                    ///merge_Candidates_list(&l, list, list_length, end_pos, 0);
+                    ///merge_Candidates_list_version(&debug_l, list, list_length, end_pos, 0);
+                }
+            }
+            else
+            {
+                avalible_k = 0;
+                init_Hash_code(&k_code);
+            }
+
+            ///HPC_base++;
+        }
+
+
+
+        ///reverse complement strand
+        reverse_complement(g_read.seq, g_read.length);
+        init_HPC_seq(&HPC_read, g_read.seq, g_read.length);
+        init_Hash_code(&k_code);
+        avalible_k = 0;
+
+        while ((code = get_HPC_code(&HPC_read, &end_pos)) != 6)
+        {
+            if(code < 4)
+            {
+                k_mer_append(&k_code,code,k_mer_length);
+                avalible_k++;
+                if (avalible_k>=k_mer_length)
+                {
+                    list_length = locate_Total_Pos_Table(&PCB, &k_code, &list, k_mer_length, &sub_ID);
+                    if (list_length != 0)
+                    {
+                        append_k_mer_pos_list_alloc(&array_list, list, list_length, end_pos, 1);
+                    }
+                    ///merge_Candidates_list(&l, list, list_length, end_pos, 1);
+                    ///merge_Candidates_list_version(&debug_l, list, list_length, end_pos, 1);
+                }
+            }
+            else
+            {
+                avalible_k = 0;
+                init_Hash_code(&k_code);
+            }
+
+            ///HPC_base++;
+        }
+
+        merge_k_mer_pos_list_alloc(&array_list, &l);
+
+        ///debug_merge_result(&l, &debug_l);
+
+    }
+    
+    destory_Candidates_list(&l);
+    ///destory_Candidates_list(&debug_l);
+
+
+    destory_k_mer_pos_list_alloc(&array_list);
+}
+
+
+
+
+
+
+
+
 void Overlap_calculate_multipe_thr()
 {
     double start_time = Get_T();

Hash_Table.cpp

@@ -24,17 +24,60 @@ void clear_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list)
 }
 
 
-void append_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list, k_mer_pos* n_list, uint64_t n_length)
+void append_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list, k_mer_pos* n_list, uint64_t n_length, 
+uint64_t n_end_pos, uint8_t n_direction)
 {
-    list->length++;
+   
-    if (list->length > list->size)
+    if (list->length + 1 > list->size)
     {
         list->size = list->size * 2;
         list->list = (k_mer_pos_list*)realloc(list->list, sizeof(k_mer_pos_list)*list->size);
     }
     
-    list->list[list->length - 1].list = n_list;
+    list->list[list->length].list = n_list;
-    list->list[list->length - 1].length = n_length;
+    list->list[list->length].length = n_length;
+    list->list[list->length].direction = n_direction;
+    list->list[list->length].end_pos = n_end_pos;
+
+    list->length++;
+}
+
+int cmp_k_mer_pos_list(const void * a, const void * b)
+{
+    if ((*(k_mer_pos_list*)a).length > (*(k_mer_pos_list*)b).length)
+    {
+        return 1;
+    }
+    else if ((*(k_mer_pos_list*)a).length < (*(k_mer_pos_list*)b).length)
+    {
+        return -1;
+    }
+    else
+    {
+        return 0;
+    }
+    
+}
+
+
+void merge_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list, Candidates_list* candidates)
+{
+    qsort(list->list, list->length, sizeof(k_mer_pos_list), cmp_k_mer_pos_list);
+
+    uint64_t i;
+    for (i = 0; i < list->length; i++)
+    {
+        /**
+        if (i > 0 && (list->list[i].length < list->list[i-1].length || list->list[i].length == 0 || list->list[i-1].length == 0))
+        {
+            fprintf(stderr, "ERROR\n");
+        }
+        **/
+        
+        merge_Candidates_list(candidates, list->list[i].list, list->list[i].length, 
+        list->list[i].end_pos, list->list[i].direction);
+    }
+    
 }
 
 void init_Count_Table(Count_Table** table)

Hash_Table.h

@@ -41,6 +41,8 @@ typedef struct
 {
     k_mer_pos* list;
     uint64_t length;
+    uint8_t direction;
+    uint64_t end_pos;
 } k_mer_pos_list;
 
 typedef struct
@@ -339,6 +341,13 @@ void destory_Candidates_list(Candidates_list* l);
 void merge_Candidates_list(Candidates_list* l, k_mer_pos* n_list, uint64_t n_lengh, uint64_t end_pos, int strand);
 
 
+void init_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list);
+void destory_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list);
+void clear_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list);
+void append_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list, k_mer_pos* n_list, uint64_t n_length, 
+uint64_t n_end_pos, uint8_t n_direction);
+void merge_k_mer_pos_list_alloc(k_mer_pos_list_alloc* list, Candidates_list* candidates);
+