ChemPup-0.1.2

I recently had the idea to assemble a Puppy Linux "Puplet" that would be configured with software useful for an undergraduate chemistry student. Considering I know nothing about programming and I am still very new to Linux, this is proving to be a challenge...but I have had success so far.

ChemPup-0.1.2 has been developed from Puppy Linux 4.21 running with 512 MB ram in a VBox machine on Ubuntu-9.04 and so far includes most of the packages I have envisioned in this first generation build. The following is the list of added packages:

Office Productivity Suite:
OpenOffice.org-3.0

General Chemistry:
GElemental - a gtk+ periodic table of the elements with loads of information on each element
MoleCalc - a molecular weight calculator (runs in WINE)

Organic Chemistry:
BKChem - a 2D structure editor with .mol and .odt (openoffice) export capabilities
OpenBabel - a command line molecular file-type converter that handles over a dozen formats
WinDNMR - a dynamic NMR simulator (runs in WINE)

Need:
A suitable 3D molecular viewer (I want Avogadro which includes a force-field minimizer)

Plan to Make:
A slick spreadsheet template (OOCalc) that has useful conversions (mole to gram etc) and equations
Detailed documentation to maximize usage of the system (since most US Students don't know Linux)

Additionally I need to figure out what programs I should cut out to save space, if any, considering OOO is adding most of the extra weight. I will probably remove devx.sfs.

Building in the existing packages (listed above) was fairly straightforward....I needed to install Python, WINE (which I found as a .pet on the Puppy Forum) and a few other package dependencies that I cannot recall.

Update: In order to run BKChem it needs python 2.3 or later and after checking with > python --version <> ./configure; make; make install. Unfortunately these programs must be installed before python so I had to reconfigure python and went ahead and downloaded the upgraded 2.6. Python is a bit large so I removed the old version...2.6 also took several minutues (10 ish) to compile and install. Afterwards bkchem.py runs from command line with > python bkchem.py < in the working directory. I then easily created a bash script to start the program, made it executable with chmod 777 and added it to the desktop with a logo icon to make a simple point and click execution.

The real trouble has been with Avogadro. This is a nice, intuitive 3D molecular viewer and force-field minimizer program that I test-ran on Ubuntu and was quite happy with, however, unlike with Ubuntu, I do not have a prepackaged .deb (.pet) for Puppy and will have to build it from source. Avogadro has several package dependencies, some of which I am having trouble with building as well: Qt4, Eigen2, CMake, and OpenBabel are all required before you can compile and install Avogadro. I installed OpenBabel and CMake okay (I think), and Eigen2 supposedly only need to be unzipped and it is ready to go (although they don't say to which directory it belongs so I stuck it in /usr/local/bin). Qt4 on the other hand has proven very difficult and I also found that I had the wrong version of CMake!! These programs (OpenBabel, Qt4, and even CMake) took a long 20-30 minutes of compiling time each. So long as I can get a user-friendly 3D molecular viewer I will be satisfied with this first release.

Maintaining Office Productivity: Virtually Solved

Currently, my "job" is school.  I am on a research assistant-ship pursuing a Ph.D. in physical organic chemistry.  Part of my degree program has involved learning the "art" of computational chemistry / molecular modeling.  Given the inherent toxicity in chemistry and my general aversion to odors, I really enjoy doing chemistry in silico.  Having made a fairly substantial leap to a linux desktop has presented a number of challenges in this regard.  First there has been the obvious struggle of losing some 'compatibility' with MS Office, but I believe this to now be a non-issue.  The real struggle is continuing to use niche chemical software: Cambridgesoft's Chem3D, ChemDraw, E-Notebook; Wavefunction's Spartan; Gaussian's Suite of Programs including GaussView; SciFinder Scholar etc.  Only Gaussian has a suitable linux version, but that has been difficult to obtain.

I have been mulling this problem over for nearly two-weeks and made several attempts at either finding a replacement software for each of the programs or otherwise attempting a "virtual" solution.  I was able to install SciFinder in WINE which is a very nice "non-emulator."  As mentioned before, I have found a few incomplete solutions for ChemDraw (BKChem, XDrawChem) and I now have tried a few different molecular GUIs including Viewmol, PyMol, Gabedit, and Tinker's Force Field Explorer (FFE).  Pymol is powerful and has great aesthetic potential as a molecular viewer, but does not serve well as a molecular editor.  Gabedit has lots of apparent functionality working as a front-end to a number of computational programs (G03, MOPAC, GAMESS, MOLPRO) but its editor is very clunky and lacks the intuitive feel I've grown accustomed to.  Tinker and FFE I expect will eventually serve as a useful way to 'clean-up' structures with a force-field calculation, but it too has no builder!  

After contacting support at PQS (a computational chemistry solutions company for hardware and software integrated in a parallel computing environment) I was given some advice to transfer the working GaussView install on the cluster computer (running Novell SUSE) to my Ubuntu 9.04 desktop.  So far, that has not been a trivial task.

So somewhat at a loss, I have swallowed my new-found linux pride and installed XP on a virtual machine created through VirtualBox 3.0.  This too required about 3 hours of research, trials and tribulations, but it seems to be the most powerful solution.  There are lots of forums on the topic of this configuration, some of which are succinct and informative.  Here are the links:

http://www.virtualbox.org/wiki/Downloads

http://download.virtualbox.org/virtualbox/3.0.0/UserManual.pdf

Supposedly it is possible to use VirtualBox to boot an existing OS installed on a partition of your HDD, but this requires a lot more understanding and caution and still has many bugs.  Otherwise the process is quite straightforward:

1. Download and install VBox for your OS-distro

2. Create a Virtual Machine using the wizard: check the VBox Forums for discussions regarding set-up, configuring your virtual disk, etc:

http://forums.virtualbox.org/

3. Install a 'guest' operating system

4. Install "guest additions" from the 'Devices' drop-down menu of your window holding the guest OS

5. Set-up shared folders to allow file access between 'host' and 'guest'

That's as far as I am for now.  I have installed most of the ChemOffice Suite (just need E-Notebook, and a key-crack) but still need GaussView and Spartan.  The system runs quite fast, but I do have 4GB RAM on my 'host' and allocated 1024MB RAM for my 'guest.'  

Betty Crocker Building Source Code

Since converting to linux (Ubuntu 9.04) at school, I have slowly been reconfiguring my chemical toolbox.  I am using SciFinder Scholar (site licence from the university) through WINE and BKChem and xdrawchem for 2D structure creation.  I still need a 3D molecular viewer and builder that would replace gaussview for windows.  This has been a bit troublsome so far.  The first challenge is simply to "test-drive" different programs available.  After reading about some of the features of a few possible programs, I settled on taking a ride in MacMolPlt 7.1.  However, just to get in the driver seat, I found myself sucked down a "rabbit hole" of dependencies.  MacMolPlt relies on wxWidgets which depends on the GTK+ 2.0 development package which in turn depends on several libraries: ATK, TIFF, glib, pango, and cairo...all of which I apparently did not have installed.  But like Alice, I was too enticed by the prospect before me and down I went.

Building from source so far seems a bit like making a birthday cake.  Since I lack the skills of both a computer scientist and a master chef, in both cases I am having to make due with getting my ingredients "from a box."  What Betty has done for millions of domestics, the Synaptic package manager does for me.  Most times I try to follow the INSTALL instructions and read the README and usually my configure step will fail.  I found all the libraries that I couldn't build from scratch using Synaptic and once installed, my configure step for wxWidgets passed.  Still this took a couple hours with lots of trial and error.  I later found wxWidgets for GTK in Synaptic also.  

In the end, I fulfilled all the dependencies for MacMolPlt but still got a strange error in the 'make' step.

The real surprise came when I went back to work and as applications were being reloaded, all the text displayed in GNOME was turning to empty boxes!!  After panicking a little, I tried to use the 'recovery' boot option and that didn't help.  After rebooting I decided to try to use my web-browser alone to back-up my files.  Fortunately I have GoDaddy server space and I was able to upload my Dissertation folder.  I couldn't back it up any other way as even the terminal window was broken.  I had older backups on a thumb drive and my other partition, but I would have lost a little more than a day of work....too much.  When I reinstalled Ubuntu 9.04 (now more experienced) I made unique partitions for /root, /usr, /opt, and /home so I won't have to worry about losing my files or programs if I "break" my installation.  I am now reinstalling all my software: Opera, BKChem, xdrawchem, Tinker, JabRef, WINE, SciFinder, Filezilla, etc and downloading my Dissertation folder from my server.  Only lost about 3 hours....

As always, lessons learned.

Writing my Dissertation: Linux or Windows?

I have noted previously that I am a Ph.D. student writing his dissertation. Recently I've been having some saving problems in MS Word and I've lost a couple paragraphs of writing twice. I have reconfigured my school desktop to dual-boot xp and ubuntu (jaunty) and I started to determine if I could accomplish writing my dissertation within linux. I already have about 40 pages of what should be around ~150 page document including over a dozen figures, schemes and tables. One of the likely culprits is Endnote, a powerful bibliography database manager which seamlessly manages your MS Office Word citations in any format you need. However, it may be inturrupting Word in some ways and causing it to flake out on me on save. And yet it is too useful to discard so I wondered if their was an alternative for Sun's OO-Writer.

There is and it is called JabRef. So far, JabRef seems just as useful as Endnote only perhaps not quite as powerful. It doesn't imbed a widget toolbar in Writer and if you delete a citation from your text it doesn't seem to omit it from the references section. (Although I still need to spend more time learning its capabilities.) [Update: JabRef works great. I imported my endnote database with ease, although I had to do some manual edits. Also, there is a "sync" button in the plugin that deals with any edits you make. The only downside (documented on Alver's site) is that you must only use the style-file downloaded directly from the web. You can edit it at will but I wouldn't even try to rename it or copy it in anyway or you are "stuck like chuck"] It can import from Endnote, SciFinder and numerous other databases with various formats. To get it to integrate with Writer a plugin (written by Morten Alver) must be added. Alver's site (see below or click here) details installation and includes a needed 'style-file' which tell JabRef how to format your citations.

When I installed JabRef from Synaptic, I found it was not the most up-to-date version. When I installed the .tar.bz2 from the JabRef site, it did not make available an icon in the menu which is a nice feature for commonly used programs (I don't mind command line, but I just like the clean feel of a menu icon to start-up). Although I had no idea how to fix this I found enough information in this ubuntu forum thread to get the job done. Very Nice!

In addition to JabRef, I will need a 2D chemical structure editor to generate structures for my figures and schemes. I love CambridgeSoft's ChemDraw and have used it freely (at school) for years. At home I had a copy of Isis Draw which was a free windows tool. In linux I am not fully satisfied here. I haven't found a single editor of nearly the capabilities as these programs. However, I did find a well written review on several available programs which can be found here. Possibly by using a few of these you can have the power of chemdraw. So far BKChem is the favorite by a small margin. Fortunately I mostly work with flat polycyclic hydrocarbons so I think BKChem will work for me.

In summary: Although it requires a little more work to install and configure and lacks marginally in functionality, at least for now, there appears to be a viable alternative for serious chemical thesis writing. I will probably make the effort to convert.

Links:

http://jabref.sourceforge.net/index.php

http://www.itk.ntnu.no/ansatte/Alver_Omholt_Morten/jabref/OOPlugin.html

http://groundstate.ca/chemdraw

http://ubuntuforums.org/showthread.php?t=225390

For an extensive list in chemical software written for linux see this link:

http://www.redbrick.dcu.ie/~noel/linux4chemistry/linux4chemistry.py