Merge branch 'dev'

ChangeLog.md

@@ -1,3 +1,9 @@
+### 0.3.3 / 2015-07-04 HKT
+
+* Fixed segmentation fault when a read is all N
+* Fixed continue mode: check continue mode before writing binary reads
+* Slightly improve SdBG traversal functions
+
 ### 0.3.2-beta / 2015-06-28 HKT
 
 * fine tune local assembly multi-thread schedules

Makefile

@@ -34,7 +34,7 @@ ifeq ($(use_gpu), 1)
 	NVCC_VERSION = $(strip $(shell nvcc --version | grep release | sed 's/.*release //' |  sed 's/,.*//'))
 endif
 
-version = $(shell git describe --tag || echo "git_not_found" 2>/dev/null)
+version = $(shell git describe --tag 2>/dev/null || echo "git_not_found" 2>/dev/null)
 
 # detect OS
 OSUPPER = $(shell uname -s 2>/dev/null | tr [:lower:] [:upper:])

atomic_bit_vector.h

@@ -37,7 +37,6 @@ class AtomicBitVector {
         capacity_ = num_words_;
         if (num_words_ != 0) {
             data_ = (word_t*) MallocAndCheck(sizeof(word_t) * num_words_, __FILE__, __LINE__);
-            assert(data_ != NULL);
             memset(data_, 0, sizeof(word_t) * num_words_);
         } else {
             data_ = NULL;
@@ -85,11 +84,7 @@ class AtomicBitVector {
         size_ = size;
         num_words_ = (size + kBitsPerWord - 1) / kBitsPerWord;
         if (capacity_ < num_words_) {
-            word_t *new_data = (word_t*) MallocAndCheck(sizeof(word_t) * num_words_, __FILE__, __LINE__);
-            assert(new_data != NULL);
-            if (data_ != NULL) {
-                free(data_);
-            }
+            word_t *new_data = (word_t*) ReAllocAndCheck(data_, sizeof(word_t) * num_words_, __FILE__, __LINE__);
             data_ = new_data;
             capacity_ = num_words_;
         }

definitions.h

@@ -26,7 +26,7 @@
 #include "kmer.h"
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.3.2-beta"
+#define PACKAGE_VERSION "v0.3.3"
 #endif
 
 typedef uint16_t multi_t;

megahit

@@ -951,11 +951,12 @@ def main(argv = None):
         logging.info("--- [%s] Start assembly. Number of CPU threads %d ---" % (datetime.now().strftime("%c"), opt.num_cpu_threads))
         logging.info("--- [%s] k list: %s ---" % (datetime.now().strftime("%c"), ','.join(map(str, opt.k_list))))
 
+        if not opt.continue_mode:
+            write_opt(argv[1:]) # for --continue
+
         write_lib()
         build_lib()
 
-        if not opt.continue_mode:
-            write_opt(argv[1:]) # for --continue
         build_first_graph()
         assemble(opt.k_min)
 

mem_file_checker-inl.h

@@ -56,9 +56,7 @@ inline void* ReAllocAndCheck(void *ptr,
                              const char *realloc_from_which_file = __FILE__,
                              int realloc_from_which_line = __LINE__) {
     void *new_ptr = realloc(ptr, size_in_byte);
-    if (size_in_byte == 0 || ptr != NULL) {
-        return new_ptr;
-    } else {
+    if (size_in_byte != 0 && new_ptr == NULL) {
         fprintf(stderr, "[ERROR] Ran out of memory while re-applying %llubytes\n", (unsigned long long)size_in_byte);
         fprintf(stderr, "In file: %s, line %d\n", realloc_from_which_file, realloc_from_which_line);
         fprintf(stderr, "There may be errors as follows:\n");
@@ -68,6 +66,8 @@ inline void* ReAllocAndCheck(void *ptr,
         free(ptr);
         exit(-1);
     }
+
+    return new_ptr;
 }
 
 #endif // MEM_FILE_CHECKER_INL_H__

sequence_package.h

@@ -273,6 +273,7 @@ struct SequencePackage {
             max_read_len_ = len;
         }
 
+        if (len == 0) { return; }
         if (len * 2 <= unused_bits_) {
             unused_bits_ -= len * 2;
             packed_seq.back() |= s[0] >> (kCharsPerWord - len) * 2 << unused_bits_;
@@ -313,6 +314,8 @@ struct SequencePackage {
     }
 
     void AddRevSeqToPackedSeq_(const word_t *rs, int len) {
+        if (len == 0) { return; }
+
         std::vector<word_t> s(rs, rs + (len + kCharsPerWord - 1) / kCharsPerWord);
 
         for (int j = 0; j < (int)s.size(); ++j) {

succinct_dbg.h

@@ -72,6 +72,10 @@ class SuccinctDBG {
         }
         rs_w_.Build(w_, size);
         rs_last_.Build(last_, size);
+
+        for (int i = 0; i < kAlphabetSize + 2; ++i) {
+            rank_f_[i] = rs_last_.Rank(f_[i] - 1);
+        }
     }
 
     uint8_t GetW(int64_t x) {
@@ -138,12 +142,12 @@ class SuccinctDBG {
             a -= 4;
         }
         int64_t count_a = rs_w_.Rank(a, x);
-        return rs_last_.Select(rs_last_.Rank(f_[a] - 1) + count_a - 1);
+        return rs_last_.Select(rank_f_[a] + count_a - 1);
     }
 
     int64_t Backward(int64_t x) { // the first node points to x
         uint8_t a = GetNodeLastChar(x);
-        int64_t count_a = rs_last_.Rank(x - 1) - rs_last_.Rank(f_[a] - 1);
+        int64_t count_a = rs_last_.Rank(x - 1) - rank_f_[a];
         return rs_w_.Select(a, count_a);
     }
 
@@ -188,6 +192,7 @@ class SuccinctDBG {
     khash_t(k64v16) *large_multi_h_;
 
     long long f_[kAlphabetSize + 2];
+    long long rank_f_[kAlphabetSize + 2]; // = rs_last_.Rank(f_[i] - 1)
 
     unsigned int num_dollar_nodes_;
     int uint32_per_dollar_nodes_;

tools/filter_by_len.cpp

@@ -32,8 +32,8 @@ int main_filter_by_len(int argc, char **argv) {
 
     long long total_bases = hist.sum();
 
-    fprintf(stderr, "%d contigs, total %lld bp, min %lld bp, max %lld bp, avg %.2lf bp, N50 %lld bp\n",
-                    (int)hist.size(), total_bases, hist.minimum(), hist.maximum(), hist.mean(), hist.Nx(total_bases * 0.5));
+    fprintf(stderr, "%d contigs, total %lld bp, min %lld bp, max %lld bp, avg %d bp, N50 %lld bp\n",
+                    (int)hist.size(), total_bases, hist.minimum(), hist.maximum(), int(hist.mean() + 0.5), hist.Nx(total_bases * 0.5));
 
     kseq_destroy(seq);
     gzclose(fp);

unitig_graph.cpp

@@ -874,7 +874,7 @@ void UnitigGraph::OutputContigs(FILE *contig_file, FILE *final_file, Histgram<in
             continue;
         }
 
-        double multi = std::min((double)kMaxMulti_t, (double)vertices_[i].depth / vertices_[i].length + 0.5);
+        double multi = std::min((double)kMaxMulti_t, (double)vertices_[i].depth / vertices_[i].length);
         if (change_only) {
             multi = 1;
         }