Pymbar can't estimate bound on free energy

[nhoutz1]

I have performed a bunch of thermodynamic-integration simulations for liquid water in Gromacs, and am trying to analyze the .xvg files. I'm using Windows 10 and installed Anaconda2 (64 bit) with the gui installer and then pymbar through Anaconda2 before downloading alchemical-analysis-master.

In a Windows command prompt, I entered:

python alchemical_analysis.py -d C:<xvg_directory>\ -t 200 -p 1 -v > results

This is the result:
Traceback (most recent call last):
File "alchemical_analysis.py", line 1224, in
main()
File "alchemical_analysis.py", line 1166, in main
nsnapshots, lv, dhdlt, u_klt = parser_gromacs.readDataGromacs(P)
File "C:<...>\alchemical-analysis-master\alchemical_analysis\parser_gromacs.py", line 187, in readDataGromacs
fs = sorted(fs, key=F.sortedHelper)
File "C:<...>\alchemical-analysis-master\alchemical_analysis\parser_gromacs.py", line 48, in sortedHelper
self.state = l[0] = int(l[0]) # Will be of use for selective MBAR analysis.
IndexError: list index out of range

where <...> is just the directory to which I downloaded alchemical-analysis-master. I don't understand the error and cannot find help for it online. In the directory where I have the .xvg files, all the files are simply named integers: 0.xvg, 1.xvg, 2.xvg, .... 20.xvg, and there are no other files in that folder. Am I calling the script correctly? Could it be the operating system (Windows vs. Linux)? I might be able to get on a Linux machine if it would help.

[davidlmobley]

Can you attach (i.e. via Dropbox or Google drive) your xvg files?

Also, it could be a naming issue - I think we're expecting xvg files to
have names with a dot notation, i.e. "name.0.xvg", "name.1.xvg", or MAYBE
'0.xvg', '1.xvg', etc. But "0xvg" and "1xvg" will probably not work.

Let us know!

On Tue, Dec 15, 2015 at 9:59 PM, nhoutz1 notifications@github.com wrote:

I have performed a bunch of thermodynamic-integration simulations for
liquid water in Gromacs, and am trying to analyze the xvg files I'm using
Windows 10 and installed Anaconda2 (64 bit) with the gui installer and then
pymbar through Anaconda2 before downloading alchemical-analysis-master

In a Windows command prompt, I entered:

python alchemical_analysispy -d C:<xvg_directory>\ -t 200 -p 1 -v > results

This is the result:
Traceback (most recent call last):
File "alchemical_analysispy", line 1224, in
main()
File "alchemical_analysispy", line 1166, in main
nsnapshots, lv, dhdlt, u_klt = parser_gromacsreadDataGromacs(P)
File
"C:<>\alchemical-analysis-master\alchemical_analysis\parser_gromacspy",
line 187, in readDataGromacs
fs = sorted(fs, key=FsortedHelper)
File
"C:<>\alchemical-analysis-master\alchemical_analysis\parser_gromacspy",
line 48, in sortedHelper
selfstate = l[0] = int(l[0]) # Will be of use for selective MBAR analysis
IndexError: list index out of range

where <> is just the directory to which I downloaded
alchemical-analysis-master I don't understand the error and cannot find
help for it online In the directory where I have the xvg files, all the
files are simply named integers: 0xvg, 1xvg, 2xvg, 20xvg, and there are no
other files in that folder Am I calling the script correctly? Could it be
the operating system (Windows vs Linux)? I might be able to get on a Linux
machine if it would help

—
Reply to this email directly or view it on GitHub
#49.

David Mobley
Associate Professor
Department of Pharmaceutical Sciences
Department of Chemistry
3134B Natural Sciences I
University of California, Irvine
Irvine, CA 92697
dmobley@uci.edu
work (949) 824-6383
cell (949) 385-2436

[nhoutz1]

Sure, I uploaded my .xvg files here: https://drive.google.com/folderview?id=0B6lGyVspwiohb2Q5eGpmRG05b3M&usp=sharing. And they do have dots in the names but I can see in your email reply the dots didn't show up for you. I tried renaming everything 'a.0.xvg', 'a.1.xvg' etc. but then the error became:

ERROR!
No files found within directory 'C:<xvg_directory>' with prefix '1' and suffix 'xvg': check your inputs.

Thinking the code might be upset by the fact that my names begin with 0 (gromacs convention) instead of 1, I tried excluding 0.xvg by moving it to another folder to see if it would run but I got the same error as before where something complains when it tries to access the 0th (1st?) index of some array.

Thank you for your help!

[davidlmobley]

No, we use numbering beginning with 0/1 as well.

Could you provide the exact command you use to alchemical_analysis when you
used the prefix "a"? I'm assuming that in the case you mentioned
("a.0.xvg", "a.1.xvg", etc.) you DID remember to use the -p option to
specify "-p a"? (The option "-p 1" in your original case is incorrect, as
this should be the prefix BEFORE the variable digits and .xvg extension.)

If so, then it's probably an issue where we need to fix something to play
nice with Windows. Once I have the command you ran, I can test under Linux
or OS X to confirm that everything works, and if it does it would mean it's
an issue with path handling under windows.

Thanks.

On Wed, Dec 16, 2015 at 5:26 PM, nhoutz1 notifications@github.com wrote:

Sure, I uploaded my .xvg files here:
https://drive.google.com/folderview?id=0B6lGyVspwiohb2Q5eGpmRG05b3M&usp=sharing.
And they do have dots in the names but I can see in your email reply the
dots didn't show up for you. I tried renaming everything 'a.0.xvg',
'a.1.xvg' etc. but then the error became:

ERROR!
No files found within directory 'C:<xvg_directory>' with prefix '1' and
suffix 'xvg': check your inputs.

Thinking the code might be upset by the fact that my names begin with 0
(gromacs convention) instead of 1, I tried excluding 0.xvg by moving it to
another folder to see if it would run but I got the same error as before
where something complains when it tries to access the 0th (1st?) index of
some array.

Thank you for your help!

—
Reply to this email directly or view it on GitHub
#49 (comment)
.

David Mobley
dmobley@gmail.com
949-385-2436

[nhoutz1]

That was the issue! I thought the -p flag was for pressure, not prefix. I think it used to be, anyway, and that is what it says in the old tutorial that I linked in my original post. I thought I was telling the program that my system was at 1 atmosphere, hence the '-p 1' which I did not change when I used the 'a.#.xvg' naming convention. Sorry, I should have read through all the flags on the main page.

Unfortunately, I now get a new error. Calling it like this:

python alchemical_analysis.py -d C:<xvg_directory>\ -t 200 -p a -v > results

Results in this:
C:\Anaconda2\lib\site-packages\pymbar\utils.py:70: FutureWarning: comparison to None will result in an elementwise object comparison in the future.
if N_k == None:
C:\Anaconda2\lib\site-packages\pymbar\utils.py:320: RuntimeWarning: invalid value encountered in log
out = np.log(numexpr.evaluate("exp(a - a_max)").sum(axis))
Traceback (most recent call last):
File "alchemical_analysis.py", line 1224, in
main()
File "alchemical_analysis.py", line 1200, in main
Deltaf_ij, dDeltaf_ij = estimatewithMBAR(u_kln, N_k, P.relative_tolerance, regular_estimate=True)
File "alchemical_analysis.py", line 282, in estimatewithMBAR
MBAR = pymbar.mbar.MBAR(u_kln, N_k, verbose = P.verbose, relative_tolerance = reltol, initialize = P.init_with)
File "C:\Anaconda2\lib\site-packages\pymbar\mbar.py", line 257, in init
self._initializeFreeEnergies(verbose, method=initialize)
File "C:\Anaconda2\lib\site-packages\pymbar\mbar.py", line 1622, in _initializeFreeEnergies
w_F, w_R, relative_tolerance=0.000001, verbose=False, compute_uncertainty=False)
File "C:\Anaconda2\lib\site-packages\pymbar\bar.py", line 317, in BAR
raise BoundsError(message)
pymbar.utils.BoundsError: WARNING: Cannot determine bound on free energy