Thank you for your fast and kind reply yesterday.
I have questioned about predicting a protein of > 1,500 aa in length.
You send me the helpful answer: http://zhanglab.ccmb.med.umich.edu/bbs/?q=node/3785
In advance, I split a long protein sequence (1,636 aa) into two fragments using two approaches.
[combination#1] 1,636 aa —> 180 aa + 1500 aa (overlapped 66 aa)
[combination#2] 1,636 aa —> 659 aa + 1500 aa (overlapped 545 aa)
Then, I have tried to use two methods, TM-score and FG-MD.
First, I performed the TM-score and got the 4 files, such as Axxxxxx.pdb, Bxxxxxx.pdb, Cxxxxxx.pdb, Dxxxxxx.pdb. I tested several combination of split fragments as follows:
[combination#1] 180 aa + 1500 aa —> superimposition
Result: http://zhanglab.ccmb.med.umich.edu/TM-score/tmp/729105.html
(See PDF 1,2 pages)
[combination#2] 659 aa + 1500 aa —> superimposition
[2-1] 653 aa (model1.pdb) was used.
Result: http://zhanglab.ccmb.med.umich.edu/TM-score/tmp/189905.html
(See PDF 3, 4 pages)
[2-2]: 653 aa (model2.pdb) was used.
Result: http://zhanglab.ccmb.med.umich.edu/TM-score/tmp/493997.html
(See PDF 3, 5 pages)
At this step, I have a question about superimposition. Each result of the TM-score vary depending on the combination of two PDB files. Could you recommend or answer what combination should be selected as the input data for accurate superimposition?
Second, I tried to use the FG-MD with D729105.pdb.
However, the server says, “ERROR, Please input a PDB formatted structure file and submit again."
So, when I look into the PDB (D729105, link to the Dropbox url) file got by I-TASSER, its column order is different from the PDB (1hdoA, see attachment) file of benchmark folder.
- Dropbox url: https://goo.gl/HquEw0
At this step, I have a question about solving the PDB format error. Would you tell me how to do easily transformation of the current pdb into adequate pdb? For example, is it possible to use chain A and B within superimposed PDB file?
Thank you so much for your professional advice.