Friday, July 29, 2011

DIYDodecad 1.01

I've made a couple of very small changes; download links have been updated.

- I've added even more detailed instructions in the README file. If you follow the instructions carefully, you will be able to run the program. I am only assuming a very basic level of computer competence (= the ability to read and follow instructions, the ability to type, the ability to install and start programs)

- I've added a new "progress" mode in addition to the "silent" and "verbose" (default). The "progress" mode shows the temporary solutions as the program is running, so they might be fun to see how the admixture proportions converge to the final solution.

A thing to note is that if the admixture proportions appear to stay the same in successive iterations, they do not necessarily stay the same. For example, a printed 0.23% in the temporary solution may be slowly changing from 0.23001% to 0.23002% etc. With the default convergence criteria it is guaranteed that the final solution will be very close to the one received if you let it run forever.

15 comments:

  1. I did 34,643 iterations in progress mode, and the proportions were about the same. Maybe you could add multithreading.

    ReplyDelete
  2. How do I turn on progress mode? Can you provide the command? Checked the Readme but did not see it. Thanks.

    ReplyDelete
  3. Nevermind, I figured it out. (Edit 'dv3.par' to substitute "progress" for "verbose", then run as usual.)

    ReplyDelete
  4. Anyone care to explain how to turn on progress mode for us "dummies"? Thanks

    ReplyDelete
  5. Anyone care to explain how to turn on progress mode for us "dummies"? Thanks


    It's in the README. And in the post above yours.

    ReplyDelete
  6. Sorry Dienekes, I'm a newbie. The README doesn't make it clear how to "edit dv3.par". I apologize if I missed anything, but any help would be greatly appreciated.

    ReplyDelete
  7. You can use any text editor you want, such as Notepad.

    ReplyDelete
  8. I got this error: ------------------------------------------------------
    -- DIY Dodecad v 1.0 ---------------------------------
    ------------------------------------------------------
    -- Copyright (c) 2011 Dienekes Pontikos --------------
    ------------------------------------------------------
    --- More information: --------------------------------
    ----- Dienekes' Anthropology Blog --------------------
    -------- http://dienekes.blogspot.com ----------------
    ----- The Dodecad Ancestry Project -------------------
    -------- http://dodecad.blogspot.com -----------------
    ------------------------------------------------------
    At line 64 of file DIYDodecad.f file: "in.exe"
    Traceback: not available, compile with -ftrace=frame or -ftrace=full
    Fortran runtime error: System can't find file
    Warning:
    execution of commando 'DIYDodecadWin dv3.par' results Status 2

    ReplyDelete
  9. Download the whole package, not just the exe file, and follow the instructions.

    ReplyDelete
  10. I downloaded the whole package, followed the instructions and made the genotype file.

    ReplyDelete
  11. Did you change the directory in the operating system command prompt to the workng directory?

    ReplyDelete
  12. I changed it in R and pasted
    system('DIYDodecadWin dv3.par')

    ReplyDelete
  13. First, thank you for sharing your software.

    My results below in the event that anyone is interested(my wife really doesn't care)

    My results would appear to be even more diverse than I had expected (no east African though).


    FINAL ADMIXTURE PROPORTIONS:
    10095 iterations
    Log Likelihood = -1.6208452339E+05
    -----------------------------
    East_European 13.91%
    West_European 39.81%
    Mediterranean 19.33%
    Neo_African 1.21%
    West_Asian 6.48%
    South_Asian 2.92%
    Northeast_Asian 7.84%
    Southeast_Asian 3.61%
    East_African 0.00%
    Southwest_Asian 2.01%
    Northwest_African 2.46%
    Palaeo_African 0.41%

    Compare this to the results that Dr McDonald had for me

    Male Most likely fit is:

    74.3% (+- 0.3%) Europe (all Western Europe)

    and 10.6% (+- 0.1%) Mideast (all North Africa)

    and 12.5% (+- 0.4%) America (all Native (Central) American) and 2.6% (+- 0.3%) E. Asia (various subcontinents)



    Any thoughts or comments are welcome.

    ReplyDelete
  14. Also, are any of these groupings other than Northeast_Asian (7.84%) commonly indicative of Amerindian heritage?

    I don't really have another explanation for Southeast_Asian (3.61%)

    ReplyDelete
  15. Amerindian heritage is not covered from this test. From past experiments, Amerindians tend to group with North Asians more than they do with South Asians, but really the results are unpredictable if a genome with substantial Amerindian ancestry is tested with 'dv3'.

    ReplyDelete