![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() | ||||||||
![]() | ||||||||
![]() |
![]() |
![]()
This document contains known problems and fixes for the SGI version of CLUSTAL W. If you have any questions about this page, please contact Dmitri Mikhailov (dmitri@sgi.com). PROBLEM: CLUSTALW 1.8 may produce incorrect alignments or dump core during the Pairwise Alignment phase. This bug is not specific to SGI platforms, but is a general bug that affects all platforms. FIX: This problem is due to a memory access bug in the original CLUSTALW 1.8 source, where the code reads beyond the bounds of an allocated array. To fix this bug, modify the pairalign.c file as follows: Replace line 214 of pairalign.c: maxscore = diff(sb1, sb2, se1-sb1+1, se2-sb2+1, (sint)0, (sint)0); with the following line: maxscore = diff(sb1-1, sb2-1, se1-sb1+1, se2-sb2+1, (sint)0, (sint)0); PROBLEM: CLUSTALW 1.74 (not 1.8!) may produce incorrect or inconsistent multiple alignments and scores, across the same or different computer platforms. This bug is not specific to SGI platforms, but is a general bug that may affect all platforms. FIX: This problem is due to a memory access bug in the original CLUSTALW 1.74 source, and it has been fixed in CLUSTALW 1.8. Specifically, depending on the input sequences used, the original CLUSTAL W code may read beyond the bounds of an allocated array, resulting in unpredictable alignments and scores. To fix this bug, modify the prfalign.c file as follows: 1. Line 444 - Substitute and use the following line: for (j=0; j<MIN(aln_len[i],prf_length1); j++) 2. Line 456 - Substitute and use the following line: for (j=0; j<MIN(aln_len[i],prf_length2); j++) Last update: 9 March 2000 | ||||||
|