Monday, November 12, 2012

Testing new version of ENMTools

Sorry for the long delay between posts/versions; my postdoctoral work took me in a direction that made it very hard to update for a while.  I'm now at The Australian National University for a new postdoc, though, and I'm hoping that I'll have a little more time to keep up with ENM Tools.  On that front, here's an updated version that fixes a few minor problems:


This update consists of two main revisions and a bit of code-tidying.  The first revision was to fix the "Resample from raster: exponential" function, which didn't work properly with scores over 1.  The second was to add some code to calculate overlaps and breadths using a different method that is suitable for larger files.  To use this method, just go to the ENMTools options and click the button for "Large file overlap/breadth".

Unfortunately at the moment there's only a perl script version available - Activestate updated the Perl Dev Kit so that it stopped working with Tkx, and I'm waiting to find out how much it's going to cost to upgrade.  If anyone out there has a working copy of Perlapp and can build Windows and OSX version, feel free.  You would have my, and everyone else's, heartfelt thanks.  We'll try to get something worked out soon regardless.

Unfortunately the issue with Mac line endings is still not fixed - I don't know if it's a Tk issue or what, but it's proving more challenging to fix than it should be.  I will keep hacking at it, though, as the line ending issue is a major pain in the ass for everyone, including me.  

If anyone out there has specific requests for the next revision, please let me know!

21 comments:

  1. Hi Dan, I am new to ENM tools and really i must admit that its a great tool in the field of species distribution or habitat modelling.I have used your tool too find out the correlation between the environment variables and subsequently i have selected some predictor variables leaving out the highly correlated ones. It has left me with a question that in Maxent also we get variables contribution and importance graph and we have to go for the most contributing ones so my question is that for which method one should go with? either selecting by correlation or by percentage contribution as in maxent?
    I have a query also kindly entertain this too. I have 900 locations of a species and i want to randomly select 100 species with resulting 9 species sets following that i can obtain 9 different results consisting 100 locations each, so that i can carry out some analysis with maxent and other niche models. how it can be done? whether it is possible with ENM tool?
    I solicit your kind help and guidance .

    ReplyDelete
  2. For your first question, I think the best thing to do is to look at the literature for your species or group of species and attempt to determine whether you can make some judgement based on previous studies of the species' physiology. When that is not possible, it is very difficult to establish which variables are most important from strictly correlative studies. In that case you're often just stuck with trying to pick a reasonable subset of those variables that are not highly correlated with each other.

    For your second question, you can either do that in Maxent (not ENMTools), or you can do it by hand in Excel.

    ReplyDelete
  3. Dear Dan, Really Nice to hear this news. I did depend on ENMTools to study niche divergence. This tools greatly expanded my ability to tackle such interesting questions. Tons of thanks.

    ReplyDelete
  4. Thanks for your quick valuable response. for my second question i want to select the species occurrence randomly and if i go with excel it will contain a bias. is it the same with MAXENT? If not can u tell me the option from where it can be done. i,e. selecting random points and making a subsets from that.
    thanks in advance

    ReplyDelete
  5. Thanks Mao!

    Vivek, you can generate random numbers in Excel and use those to sort. You can also do the k-fold validation in Maxent, which I think will do what you want. I don't know the procedure for that off the top of my head, though, so I suggest you check the help that comes with Maxent.

    ReplyDelete
  6. Hi Dan, I´m new using this program. I run the identity test with 1.4 ENMTools. The I and D results in the .csv does not appears in two columns, in fact only one row appears with 200 numbers (I guess 100 for the I column and 100 for the D column). Do you know how to know which numbers are for I and which ones are for D? It was my mistake in the process of the test? Thanks in advance!!!

    ReplyDelete
  7. Are you on a Mac or a PC? This may be an issue with line endings - I was trying to fix a Mac line ending bug in this version, but it ended up not working out. It's possible that I did something in the process that is messing up your outputs.

    ReplyDelete
  8. Hi Dan, I´m using a PC with 64 bits. So it´s not possible to know which number is for each column? Any way I´m running again with the 1.3 version and its working good. Thanks for your quick response!!!

    ReplyDelete
  9. If I can diagnose the problem, it will probably be clear what numbers belong where. Gotta figure that first bit out first, though.

    ReplyDelete
  10. At present I'm not able to duplicate the problem. Is it your sorted file from the identity test that's messed up, or is it the unsorted one? Or both?

    ReplyDelete
  11. The unsorted file has the problem, and in the sorted file only appears one number in each column.

    ReplyDelete
  12. The identity test with 1.3 version finished and also has problems. Just one of the three comparisons has results, the other two comparisons files are in blank (the sorted and unsorted). I´m using the pearl for 64 bits and the maxent 333e version. I modified the RAM for Java and for Maxent to be faster. I have no idea whats is wrong with the procces or if I did something wrong with in the instalation. Can you recomend me something?

    ReplyDelete
  13. Hmmm. Can you emails me your input .csv files?

    ReplyDelete
  14. Hi Dan

    I have been successful in attaining results of the niche overlap and niche width with my data. However while using the Identity test, i have used separate .csv occurrence file for different species, but the resulting excel sheet gives blank results.

    Please guide me over this

    Best regards,
    Utsav Bahuguna

    ReplyDelete
  15. It was very nice article and it is very useful to Linux learners.We also provide Linux online training

    ReplyDelete
  16. This comment has been removed by the author.

    ReplyDelete
  17. So... Not sure this is the right place, but here it goes. I am new to ENMTools and ran a niche overlap on large files. I got the I and D as outputs, but not the RR. Am I doing something wrong or is RR not supported for large files?

    ReplyDelete
  18. Hello,
    I'm trying to install ENMtools on my windows 10 computer and in the manual it says "No installation is required, just uncompress the file provided and double click the appropriate executable (“ENMTools_3-5-2011.exe” for Windows users)" I realize this is probably a dumb question but I uncompressed the file and there are no .exe files, just .DS_Store, enmtools.config, ENMTools_1.4.4.pl, the manual, a text file, and sample data.

    ReplyDelete
    Replies
    1. After a certain point several years ago, ActiveState changed their tools so that I could no longer compile an exe version of ENMTools. Now you have to install ActiveState Perl in order to run it from the .pl file. However, I would suggest that you use the new R version of ENMTools instead if you can - it does many more things than the original ENMTools, and for many things is much faster. There's a vignette about how to download and use it here:

      https://github.com/danlwarren/ENMTools

      Delete