problems using multiple processors on V1.13

I’m currently experiencing some problems running analyses on multiple processors. So far, with both align and chimera slayer, when I run them on multiple processors not all of the temp files are added to the final data output. The temp files remain in the output directory and the total number of sequences within these files sum to the total expected, but the final output file (concatenated temp files?) only contains the data from n-1 of the temp files. e.g., a recent run of chimera.slayer returned temp files containing 447, 590, 344 and 450 sequence names. The final accnos file, however, contained only 1384. The sequence names from the first temp file were omitted. The same occurs for align.seqs and occurs for various processor numbers (i.e, if I use 2 processors to align about half of the seqs are not present in the output file, but all a present in the 2 temp. files).

I’m running mothur on 64-bit ubuntu.

Thanks for your help,


It seems that the problem here is also somehow related to the ~ sign being used in path names. When I replace the ~ with the full path the multiple processors option works successfully to completion, when I use the ~ in the path name the bug described above prevails. See also my post “dist.seqs outputs files aren’t output”. Again this seems strange, as with a single processor the tasks complete and the out files are written successfully when the ~ is used.

It seems safest, therefore, to use full path names.