2

简短版本：为什么循环迭代(compounds+k)->spectra->peaks后会发生变化for（约 2500 个化合物中有 4 个）？

长版：我有一个功能可以检查我的所有数据（在结构色谱图 ( chrom) 中）并将那些type == 1直接添加到新结构 ( compounds) 中，type == 2如果剩余数据中存在高度相似的色谱图并求和/平均，则检查色谱图进入 1 个条目compounds。当我最初使用当时拥有的数据集编写它时，该程序运行良好，但最近的数据我遇到了一个错误，其中跟踪频谱中有多少“值”的整数在包含for loop结束后以某种方式重置为 0。我希望在阅读我的代码后会更清楚一点（特别注意最后证明问题的 2 个打印件）。

chromatogram*
spectral_matcher(chromatogram* chrom, arguments* args) {
   int i, j, k = 0, l, m, size, counter = 0;
   float low_mass, high_mass, low_time, high_time;
   chromatogram* compounds;
   compounds = calloc(MAX_SPECTRA,sizeof(chromatogram));
   for (i = 0; i < chrom->hits-1; i++) {
       if ( (chrom+i)->type == 1) {
           /* Adding the MS1 spectrum to output set */
           chrom_copy(chrom, compounds, i, k);
           k++;
       } else if ((chrom+i)->used != 1 && (chrom+i)->type == 2) {
           /* Adding the MSn spectrum to output set */
           chrom_copy(chrom, compounds, i, k);
           /* Acquiring search paramenters */
           low_mass = (chrom+i)->precursor - args->mass_tolerance;
           high_mass = (chrom+i)->precursor + args->mass_tolerance;
           low_time = (chrom+i)->time - args->time_tolerance;
           high_time = (chrom+i)->time + args->time_tolerance;
           /* Performing search for matching spectra */
           for (j = i+1; j < chrom->hits; j++) {
               if ( (chrom+j)->type == 2 && (chrom+j)->precursor > low_mass && (chrom+j)->precursor < high_mass && (chrom+j)->time > low_time && (chrom+j)->time < high_time && (chrom+i)->spectra->peaks > 10 && (chrom+j)->spectra->peaks > 10 && (chrom+j)->used != 1) {
                   /* the KS test can only be performed if the previous if statement was true */
                   if (pdf_ks((chrom+i)->pdf,(chrom+j)->pdf, 1.0) == 1) {
                       if (args->verbose == 1) {
                           printf("Matching spectrum %i with %i into %i\n",i, j, k);
                       }
                       // De magicks - Photo Finish
                       counter++;
                       l = (compounds+k)->spectra->peaks;
                       size = (compounds+k)->spectra->peaks + (chrom+j)->spectra->peaks;
                       (compounds+k)->spectra->peaks = size;
                       m = 0;
                       /* `l` is at the end of original spectra, append values starting from `l` */
                       for (; l < size; l++) {
                           ((compounds+k)->spectra+l)->mz_value = ((chrom+j)->spectra+m)->mz_value;
                           ((compounds+k)->spectra+l)->int_value = ((chrom+j)->spectra+m)->int_value;
                           m++;
                       }
                       // set the 'matched' spectrum to NULL so there will be no duplicates
                       (chrom+j)->used = 1;
                   }
               }
           }
           k++;
       }
       /* k was incremented in either the if or else if so doing -1 here */
       printf("%i: values [ %i ] contains a value set [ %f - %f ]\n", k-1, (compounds+k-1)->spectra->peaks, ((compounds+k-1)->spectra+5000)->mz_value,((compounds+k-1)->spectra+5000)->int_value);
   }
   printf("BREAKPOINT\n");
   printf("%i spectra summed\n",counter);
   compounds->hits = k;
   for (i = 0; i < compounds->hits; i++) {
       printf("%i: values [ %i ] contains a value set [ %f - %f ]\n", i, (compounds+i)->spectra->peaks, ((compounds+i)->spectra+5000)->mz_value,((compounds+i)->spectra+5000)->int_value);
   }
   exit(0);
   return(compounds);
}

我知道 4 种化合物会产生我之前解释过的奇怪行为，所以这里是输出中的匹配行：

736: values [ 16481 ] contains a value set [ 765.000000 - 0.000000 ]
847: values [ 16481 ] contains a value set [ 765.000000 - 5843.000000 ]
1810: values [ 16481 ] contains a value set [ 765.000000 - 0.000000 ]
2212: values [ 16481 ] contains a value set [ 765.000000 - 0.000000 ]
BREAKPOINT
736: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]
847: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]
1810: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]
2212: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]

数组中的值甚至似乎会根据这个结果发生变化。

然而，4 个特殊值周围的值仍然正确：

735: values [ 44801 ] contains a value set [ 556.250000 - 0.000000 ]
736: values [ 16481 ] contains a value set [ 765.000000 - 0.000000 ]
737: values [ 131848 ] contains a value set [ 765.000000 - 0.000000 ]
BREAKPOINT
735: values [ 44801 ] contains a value set [ 556.250000 - 0.000000 ]
736: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]
737: values [ 131848 ] contains a value set [ 765.000000 - 0.000000 ]

如果有人对接下来要检查的内容有任何点击或提示，我将不胜感激。

-- 5 月 16 日 (4:20) --

i我尝试通过添加if (i == 806) { break; }到代码中手动打破特定值的 for 循环来查看数据更改的位置。这产生了：

736: values [ 16481 ] contains a value set [ 765.000000 - 0.000000 ]
BREAKPOINT
736: values [ 0 ] contains a value set [ 765.000000 - 905.625000 ]

——5月17日——

我还检查了 i 和 k 计数器是否在做一些奇怪的事情，但它们看起来非常好（在 for 循环内）：

I: 803  K: 734
I: 804  K: 735
I: 805  K: 736 /* The iteration which shows wrong data AFTER the for loop closes */
I: 806  K: 737
I: 807  K: 738
I: 808  K: 739
4

3 回答 3

2

看看这个：

chromatogram* compounds;
compounds = calloc(MAX_SPECTRA,sizeof(compound));

你也许是说sizeof(chromatogram)？或者compound *compounds？

于 2013-05-16T11:56:03.097 回答
0

会不会是你改变了 的值(compounds+k)->spectra->peaks？

size = (compounds+k)->spectra->peaks + (chrom+j)->spectra->peaks;
(compounds+k)->spectra->peaks = size;

size变量是 an ，这int可能与您的问题有关。

于 2013-05-16T15:59:08.997 回答
0

我正在追加分配给光谱的空间（在照片完成注释的部分中），我本来希望 valgrind 给出一条写过去缓冲区的消息，但我得到的错误后来没有无效指针。

通过增加值来解决问题MAX_PEAKS

于 2013-05-17T11:53:46.550 回答