Basic guides to APBS on PyMOL

The following instruction guides you on how to 1. Install and then 2. Run the APBS (Adaptive Poisson-Boltzmann Solver) plugin for PyMOL under Windows.

Installation

Run apbs_i386_setup.exe
The installation process will create the folder apbs immediately under the C:\ drive. You can change the installation directory, but if you choose to do so, make sure that there is no space anywhere in the directory path.

e.g.
OK  C:\apbs
BAD C:\Users\abc\path with gaps
OK  C:\Users\abc\path_without_gaps

Running the APBS

1. Prepare the PQR file from the PDB file of the structure that you want to draw the electrostatic surface potential diagrams of.
   Visit http://nbcr-222.ucsd.edu/pdb2pqr_1.8/
   
   Either input the PDB accession code, or upload a PDB file.
   
   Download the PQR file.
   
2. Launch PyMOL. Firstly, load the coordinates of the structure (.pdb file). Access Plugin -> APBS Tools 2.1.
   
   
   
   
3. Again, make sure that there is not a single gap in any of the directory paths. I imagine APBS was originally developed to run of Linux-type operating systems.
   
   
   
   
   
4. After waiting, the screen should automatically change to the “Visualisation” tab.
   
   
   
5. Voila!
   
   
   
   One can ray trace it but it doesn’t seem to make much difference to be honest (below).
   
   
   
   
   
   
   
6. Note: While running the above procedure for a larger protein (55 kDa) under Windows 8.1 (64 bit), the PC hung with an error message, “computer has stopped responding” or something like that. I feel this is non-sense, because I saw in the task manager that APBS was happily doing its calculation, just taking some time. Besides, I successfully managed to carry out under Windows 7 Home (64bit). So I think it’s Windows 8.1 is mistaking a long calculation to PC being crashed. Reading the following page (http://www.tomshardware.com/forum/61690-63-program-responding-timeout-edit) did not help, unless I missed something.
   
   Update : Using free academic PyMOL 1.3 and APBS 1.2 can successfully complete the process for the 55kDa structure that failed in the above-mentioned setup, which was PyMOL 1.6 with APBS 1.4. Also failed with PyMOL 1.7 (64bit) with APBS 1.4 (32bit or 64bit). All tested under Windows 8.1 Pro 64bit.

メインPCアップグレードの巻

しばらく使っているこのメインデスクトップPC。スペック的にはほぼ満足してるのですが、World of Tanksを最高設定でFullHD解像度で動かすと、FPSが20程度までに落ち込むのでグラボをアップグレードすることにしました。

元のスペックは大体以下の通り。

Core i5 2500K @ 4.5 GHz
DDR3 PC3-10700 8 GB
Radeon HD5770 1 GHz
Crucial M500 SSD 480 GB

前世代のミドルローであったHD5770を新世代のミドルハイであるR9 270Xに交換する魂胆です！

作業開始。真ん中に鎮座しているのか今まで頑張ってきたHD5770。

取り出して比べてみます。上がR9 270X、下がHD5770。さすがにグレードの差がサイズに表れています。

そして装着。HD5770の補助電源は6ピンx1でしたが、R9 270Xは6ピンx2です。一応電源は足りているハズですが、スイッチを入れた瞬間に発火しないのを祈ります。

ドライバをアップデートし、作業は完了。所有時間は約30分。得た嬉しさはPRICELESS。

さて、一体性能は上がったのでしょうか？


まずはエクスペリエンス・インデックス。確かにグラフィック関係は上がりましたが、劇的ではありません。

次はCinebench R11.5のOpenGL。OpenGLは個人的にあまり重視していませんが。
12.24％の伸びです。やはり、劇的ではないのか・・・



次はFFXIVキャラベンチ。これは換装前のHD5770。

そしてこれが換装後のR9 270X。妥当なスコアアップというところでしょうか。

ベンチマークを見る限りでは効果が少ないものもありますが、World of Tanksでは、FPSが以前の20前後から、50-65まで向上し、落ち込みもありません。（もちろん設定は同じです）そのスムース具合はまさに別ゲー。２倍以上のFPSの伸びが見込めるというのはレビューなどから予測していましたが、それを体感するのは別の話。買ってよかったAMD！

今まで使ってきたグラボ歴を綴ると、

• RIVA TNT2
• GeForce FX5500
• GeForce 6600GT
• Radeon HD2600XT GDDR3版
• Radeon HD5770

そして今回の

• Radeon R9 270X

となるわけです。 それぞれ世代が離れているせいか、性能の向上度を毎回すごく感じます。一番感動したのは6600GTから2600XTに変えた時のような気がします。

Espript : Trouble shooting

I like Espript very much, but it can be confusing when error messages follow your job submissions, and you don't know what you have done wrong. I document my recent experience below so that I may return to this and others may also find it useful.

1. Make sure the sequences in the alignment and PDB are consistent.

Espript looks at both aligned sequences and the PDB files, therefore it will be confused if there is an inconsistency between the sequence in the alignment and the PDB.
Often, the sequences downloaded from RCSB/NCBI are not identical to that in the PDB structure files. This is because not all coordinates can be assigned where electron densities are weak.

In this case, you can use this to extract the amino acid sequence from PDB files. And then you can use it to create a sequence alignment.

Usually this will give you an alignment that will work well for Espript.

If you are still having a problem, read on.

2. Check the quality of the PDB.

Open the "Server logs" tab in the "Results" window. Many clues can be found here if something goes wrong. In my case, the log told me it found this strange "X" amino acid in the PDB I uploaded.

On closer inspection, it turns out that the coordinates of sidechains were missing from some amino acids.
(This is a fairly common crystallographic problem. Sometimes electron densities are too weak for assigning coordinates. Actually, I like the authors' honesty for not assigning coordinates if there are no electron density)

Espript uses DSSP for extracting secondary structure information from PDBs. You can obtain the DSSP file of your PDB, and mine looked like below.

**Tip: I prefer to rename my DSSP file to something like "file.dssp" for clarity.

I "corrected" the DSSP output file by replacing those "X" amino acids by what really are suppopsed to be in those places. So in the case above, the first two "X" were replaced by K and R. Additionally, I had to make sure that the residue numbers are completely contiguous (continuous) in the "RESIDUE" column.

So now you have an edited DSSP file that describes the secondary structure of your PDB file. You can upload this DSSP file to Espript instead of the PDB file, and Espript will work.

These are what I had to go through, and hopefully it may solve your trouble too. Comments are appreciated. Good luck!