I have performed series of modelling experiments using modeller to get a series of modeller proteins. But here i am unable to produce models with good GA341 score. I have a hundred percent match to the query.how to improve their score of modeller. Plus suggest a better program in modeller or any other better way..by the way my current highest score is just 0.4 ( GA341 score)
You can also check the DOPE score obtained, take the structure with the lowest DOPE score. Again ONLY these criteria would not be appropriate. It should be followed by structural analysis of the modeled structure. Hope this helps.
Regards
Please understand my question. I will repeat my self. I have 100%match in BLAST and now have modelled the query sequence with the template sequence but ended up having very low GA341 score and high DOPE scores. Can any one suggest me a method to get a better model with higher GA341 score ( almost 1) and lower DOPE score.
Firstly check the quality of the template structure. Although, it should be fine but some of the older structures may not be of sufficient quality. This apart, you can try energy minimization methods as a start. Even better, if you can run short simulations using any Molecular Dynamics package. Most often than not, this will increase model quality by a notch. Finally, you can the SAVES server to check your model quality. It has multiple programs for checking and validating structures. I have provided the link below. Hope this helps.
http://services.mbi.ucla.edu/SAVES/
Depending only on the GA341 or DOPE score is not enough to come to a suitable conclusion. Is BLAST giving you 100% Identity or Similarity? Have you tried aligning the query and the template, both in terms of sequence as well as structure. If that gives you favorable results then you can select a structure without looking for a single number score. I would also advise to use other homology modeling tools by which you will get a more clear idea if different algorithms are giving you different structures or everything just gives you the same output.
You can try using multiple templates to build the model and try to use structural descriptor desides the "energetics scores" like the psi-phi values, RMSD/F, etc
Hello
If your blast identity is 100% then in my view u dont need modelling but check sequence with template by alignment. Also check DOPE score and take the lowest score. Also try some other software or servers for modelling. best server in my opinion is iTASSER
http://zhanglab.ccmb.med.umich.edu/I-TASSER/
Hello
1. Check the quality of you template,: Procheck, SAVES, ...
2. Check sequence identity between the query and the template (PDB sequence!)
3. Check your alignement in terms of secondary structure, to be sure that there no shifting.
4. Run modeller
5. Check, DOPE SCORE, GA341 and the RMSD between the template and the model. How many atoms are align?
Trust this help!
Aurore
Since this thread is still open and the QA have not confirmed, I would add my 2 cents which can help.
Dear Sheshank,
After reading your question and the comment, I have an impression that you are using the BLAST %identity to quote the identity amongst template and sequence. BLAST performs local sequence alignment, i.e. it will try to find islands of high sequence similarity between two sequences. However, the alignment required for MODELLER is a global alignment. Therefore:
1. The %identity of blast should be considered along with Query coverage and e-value.
2. Perform a global alignment between template and query using MAFFT or clustalO, promals3D, etc. to see if there are In/Dels in the query.
3. The GA341 score assesses the model reliability but takes into account the sequence identity between template and query (global alignment). If the % identity is 100%, still GA341 can be less than 1 but should not be less than 0.7
You can check the sequence identity taken into account for GA341 calculation in the log, under the tag
I hope it helps.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling