[Feature-users] Re: Evan MAsutani IRTA update

[Evan Masutani]

Hi Michael,

Since dssp-2.2.1 looks like it might be an issue, there were a couple of
other tricky things I had to do to get it to work.  First, per the README,
you need to manually enter ~/dssp-2.2.1/src and modify mkdssp.cpp and
replace all instances of "Version" (I can't remember if it was all caps) to
"2.2.1" otherwise it won't work.  I am a complete newbie with C++/C so I
had no idea how to get the thing to work, but fortunately: (
http://macosxpostdoc.blogspot.com/2012/09/two-for-price-of-one-mac-os-x-without.html)
supplied me with the command:

*c++ -O3 -o dssp -I/sw/include -L/sw/lib  -Wall
-Wno-multichar  -lboost_thread-mt -lboost_regex-mt  -lboost_filesystem-mt
-lboost_program_options-mt  -lboost_date_time-mt -lboost_iostreams-mt
-lboost_system-mt *.cpp *
It gave me some warning about an unused variable kSSBridgeDIstance and not
finding a directory for -L/sw/lib.  Other than that, it seemed to work
fine, but was oddly clunky to use.  Perhaps as a result of the above error,
I could not convert .pdb files to .dssp files unless the .pdb file was in
~/dssp-2.2.1/src (and needed to include said directory in terminal when
giving the conversion command).  I feel like this is just an error on my
part and isn't a big deal for me.  Thank you very much for your reply and
please let me know if you have any questions!

Sincerely,
Evan Masutani

On Fri, Oct 10, 2014 at 3:33 PM, Michael Wong <mikewong at sfsu.edu> wrote:

>  Hi Evan,
>
>  Hi Evan,
>
>  I’ll update the DSSP links in the documentation; previous versions of
> DSSP did *not* need the Boost libraries, so this information is very
> helpful.
>
>  DSSP v2.2.1 gives different results than DSSP v1, so the scores will be
> different. I’ve confirmed that FEATURE-3.1 gives consistent results with
> the old version of DSSP, so I’ll have to look more closely into what the
> differences are and how to manage the change. The most logical choice would
> be to update to the latest version of DSSP for our nightly build tests and
> documentation, although I’ll have to balance that with the impact on users
> who have the old DSSP.
>
>  As a bioengineer, you know that software “bit rot” is a real issue.
> FEATURE has several dependences, so whenever a dependence changes, it
> always leads to a fire-fighting adventure. Thanks for your patience!
>
>  Best regards,
>
>  - m.
>
>  __________________________________
> Mike Wong, Staff Research Associate
> Center for Computing for Life Sciences (CCLS)
> San Francisco State University
> 1600 Holloway Avenue, Hensill Hall 302, SF, CA 94132
> (415) 405-2119
> mikewong at sfsu.edu
>
>  On Oct 10, 2014, at 10:36 AM, Evan Masutani <evmasuta at gmail.com> wrote:
>
>       Dear Michael and simtk.org,
>
> Thank you very much for your reply!  I currently have feature-3.1.0
> running and it's great!  Installing it, however, had a few kinks that I
> thought might be worth asking about.  As a bit of background, I am a
> bioengineer by training and am not all that familiar with Macs (my work
> computer is a Mac running OS X 10.9.5) and am a mediocre programmer at
> best, so my apologies if I say something dumb.  For ease of reading, I have
> put the issues I encountered into a list below:
>  1) The dssp link in the manual is broken.  After the fact, I might've
> seen the dssp program in the feature folder but was not sure.  What I ended
> up doing was downloading dssp-2.2.1
>  2) The Mac I am working on did not have Boost installed and it was
> necessary to run dssp-2.2.1; using Macports resulted in some downstream
> errors, but using Homebrew worked well.
>  3) There were some other peculiarities with dssp-2.2.1 but I realize that
> this is outside the scope of FEATURE, but will be happy to give more input
> if you would like.
>  4) For whatever reason, only standard FEATURE could install on my Mac and
> it might be because I am running clang?  I am not sure, but it works fine
> otherwise.
>  5) make install feature does not add the examples folder to the
> directory; only README, bin, data, and tools are there
>  6) After fiddling around I (if I remember correctly, but not really
> relevant) managed to try the serine protease example by changing
> directories to a separate feature-3.1.0 folder on my desktop.  When I
> looked at the output, however, they differed from the output given.  For
> one, curiously, FEATURE only returned the top 2 scoring hits and I double
> checked the command in the terminal.  When I opened the .hits.sorted file,
> there were the same number of entries as with the example output.  Even
> more curiously, however, the output values themselves did not match.  I
> have pasted the outputs below:
>  Manual Output (omitted other 3 entries):
>  Env_1bqy_31 149.060025 35.096 9.444 145.234 # SER195:B at OG
>  Env_1bqy_14 176.033218 19.642 39.631 59.232 # SER195:A at OG
>  My Output:
> Env_1bqy_31 144.728524 35.096 9.444 145.234 # SER195:B at OG
> Env_1bqy_14 154.566672 19.642 39.631 59.232 # SER195:A at OG
>
>  To the best of my knowledge, I followed the directions properly, so this
> result is puzzling.  I was wondering if it had to do with the fact that I
> am using a different version of dssp?
>
>  Thank you very much for your time!  I really appreciate it and apologize
> for any ignorant comments.  Have a great weekend!
>
>  Sincerely,
> Evan Masutani
>
> On Thu, Oct 9, 2014 at 2:57 PM, Michael Wong <mikewong at sfsu.edu> wrote:
>
>> Hi Evan,
>>
>>  Thanks for using the FEATURE software! Your feedback helps improve
>> FEATURE as a solid basis for computational protein research.
>>
>>  You can post to feature-users at simtk.org and/or directly e-mail me your
>> suggestions to improve the installation process; I suggest that you e-mail
>> feature-users at simtk.org and cc me (I will get both e-mails) so we can
>> track and resolve the issues you raise.
>>
>>  Best regards,
>>
>>  - m.
>>  __________________________________
>> Mike Wong, Staff Research Associate
>> Center for Computing for Life Sciences (CCLS)
>> San Francisco State University
>> 1600 Holloway Avenue, Hensill Hall 302, SF, CA 94132
>> (415) 405-2119
>> mikewong at sfsu.edu
>>
>>  On Oct 8, 2014, at 7:15 PM, Russ B Altman <russ.altman at stanford.edu>
>> wrote:
>>
>> You should be able to post things on the discussion part of simtk.org
>> but I am cc’ing Mike Wong who helps with managing the code base, in case he
>> has a better place for you to put your info.  I’m also cc’ing a grad
>> student in my lab, Weizhuang, who has built a Ca++ site recently and may be
>> useful to you for advice.
>>
>> Thanks!
>> Russ
>>
>>
>> On Oct 7, 2014, at 4:12 PM, Evan Masutani <evmasuta at gmail.com> wrote:
>>
>> Dear Prof. Altman,
>>
>> Thank you very much for your encouraging email!  I am going to give it a
>> shot and build a model for Mg++ sites.  I am still getting the
>> software/dependencies set up, so I will play around with FEATURE for a bit
>> before soliciting further advice.  I am running it on a Mac and had some
>> difficulties as well as work-arounds with the installation and was
>> wondering if/where I should post this information.  In particular, getting
>> dssp was rather painful, although I did manage to do so in the end.  Thank
>> you very much!
>>
>> Sincerely,
>> Evan Masutani
>>
>> On Fri, Oct 3, 2014 at 6:17 PM, Russ B Altman <russ.altman at stanford.edu>
>> wrote:
>> Evan,
>>
>> Nice to hear from you!  Very interesting--yes we have a MG++ finder for
>> RNA, but have never built one for proteins.   This is chiefly because Ca++
>> and Mg++ are very very similar and we are not sure that we have good gold
>> standards because depending on which of them is in the buffer when a
>> protein is crystallized they can often occupy each other's sites.  So--you
>> are right:  the Ca++ models can sometimes be good Mg++ finders.   I am
>> going to cc a BioE grad student in my lab who has just built the definitive
>> Ca++ model, and used Mg++ as a test for specificity.  He may have ideas (or
>> code) for building a MG++ specific model that might be fun--if you are
>> interested.  I'm not sure he wants to do this because the Ca++ model was
>> just a side project and not his main PhD project, but I will let him
>> comment.
>>
>> Building FEATURE models not that hard:  you find N good Mg++ sites that
>> you trust and M good "non-sites" (Weizhuang is expert at this) and then you
>> run FEATURE to get a model, cross-validate it, and then go to town!  All
>> the FEATURE code is available at simtk.org
>>
>>
>> Thanks!
>> Russ
>>
>>
>> On Oct 3, 2014, at 1:01 PM, Evan Masutani <evmasuta at gmail.com> wrote:
>>
>> Dear Prof. Altman,
>>
>> I hope that you are well and that the new academic year is off to a good
>> start!  I am writing to inform you that all is well at my postbac position
>> at the NIH.  I will be working with magnesium transporters; as it turns
>> out, as part of the lab's projects, we are trying to identify magnesium
>> binding sites on some proteins.  The first thing I thought of when I heard
>> that was FEATURE.  I decided to try and take a look at webFEATURE and ran a
>> few proteins.  I noticed that the database lacks model proteins for
>> magnesium binding sites; however, when I was looking at the results of
>> H-Ras (which has magnesium binding sites), I saw that there were a number
>> of predicted Ca2+ and ferritin/iron binding sites.  If you have time, I
>> would really appreciate any advice on how I should go about looking for
>> magnesium binding sites bioinformatically, as well as your opinion about
>> possible crossover between predicted Ca2+/ferritin binding sites and Mg2+
>> binding sites.  Thank you very much and have a great weekend!
>>
>> Sincerely,
>> Evan Masutani
>>
>>
>>
>>
>>
>>
>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://simtk.org/pipermail/feature-users/attachments/20141010/6d660d3c/attachment-0001.html