A few bugs

mothur bugs
1

First of all, I can’t seem to download the newest version of mothur, so some of the bugs listed here may have already been fixed. When I click on the download link, the folder contained in the zip file is named mothur-1.5.0, and the compiled executable file says it’s mothur v.1.4.1. All the files in the folder are dated Aug 10. Am I doing something wrong? I’ve tried this with two broswers and emptied their cache.

Anyway, here are the bugs:

  1. libshuff only works with a phylip-formatted distance matrix. If I read in a matrix file with this command: read.dist(column=xxx.unique.aligned.dist, name=xxx.names, group=xxx.groups) and run libshuff(), it tells me “You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command.”. I have to read in a phylip-formatted matrix with this command: read.dist(phylip=xxx.unique.aligned.phylip.dist, group=xxx.groups) for it to work. This not only means I have to generate a separate phylip-formatted matrix, but also that the information contained in xxx.names (the original fasta file was deconvoluted, hence the use of column-formatted matrices) was not used by libshuff (maybe it doesn’t matter?).

  2. I often get a Segmentation fault when running get.oturep() with even a small data set. I cannot tell what’s causing the problem. Let me know if you want an example dist file that causes this problem.

  3. get.oturep only works with a column-formatted matrix. This would be less of a problem if libshuff above works correctly with column-formatted matrices.

  4. The number of distances read from phylip-formatted matrix appears to be incorrectly reported (assuming that the number from the same matrix in column-format is correct), and the numbers differ depending on whether a groups file is used.

  5. SVG files are broken. The correct header should be like this:

  6. Please include a function to batch convert SVG files to PNG files. I understand all the advantages provided by SVG, but programs that can correctly modify SVG files are not very common (AFAIK just Illustrator and InkScape).

Thanks for a great program. Please let me know if there’s anything else I can do to help.

2

First of all, I can’t seem to download the newest version of mothur, so some of the bugs listed here may have already been fixed. When I click on the download link, the folder contained in the zip file is named mothur-1.5.0, and the compiled executable file says it’s mothur v.1.4.1. All the files in the folder are dated Aug 10. Am I doing something wrong? I’ve tried this with two broswers and emptied their cache.

If the version you are using produces a logfile, you are using v1.5. We’re not big on eye candy. mothur v1.6 will hopefully be out in the week of 10/5/2009

  1. libshuff only works with a phylip-formatted distance matrix. If I read in a matrix file with this command: read.dist(column=xxx.unique.aligned.dist, name=xxx.names, group=xxx.groups) and run libshuff(), it tells me “You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command.”. I have to read in a phylip-formatted matrix with this command: read.dist(phylip=xxx.unique.aligned.phylip.dist, group=xxx.groups) for it to work. This not only means I have to generate a separate phylip-formatted matrix, but also that the information contained in xxx.names (the original fasta file was deconvoluted, hence the use of column-formatted matrices) was not used by libshuff (maybe it doesn’t matter?).

Yes. It only takes in a phylip-formatted file. Yes, the redundant names probably matter. Making slibshuff (and parsimony, unifrac, et al.) compatible with a column matrix/names file is in the list of things to do, but I have to admit it is pretty far down that list. These hypothesis testing procedures only tell you whether your libraries are different. If you have a large number of sequences, the test will probably yield a low p-value.

  1. I often get a Segmentation fault when running get.oturep() with even a small data set. I cannot tell what’s causing the problem. Let me know if you want an example dist file that causes this problem.
  1. get.oturep only works with a column-formatted matrix. This would be less of a problem if libshuff above works correctly with column-formatted matrices.

Could you email me the data inputs and we can take a look.

  1. The number of distances read from phylip-formatted matrix appears to be incorrectly reported (assuming that the number from the same matrix in column-format is correct), and the numbers differ depending on whether a groups file is used.

Can you email me an example of what you mean?

  1. SVG files are broken. The correct header should be like this:
  1. Please include a function to batch convert SVG files to PNG files. I understand all the advantages provided by SVG, but programs that can correctly modify SVG files are not very common (AFAIK just Illustrator and InkScape).

This has been fixed for 1.6. We’ll look into the PNG conversion. I’m curious what software is able to manipulate a PNG file.

Thanks for the post,
Pat