Talk:Homology-based structure prediction (PKU)

From Bioinformatikpedia

This is just a place to add criticism and or positive feedback so we Jonathan and Sebastian have a possibility to improve. Please note, that we wont be able to adress all matters immediately but we will address them eventually!

schematic group of people coloured in several colours with speech bubbles above their head
source: [1]

We also want to mention our slogan for this current teamwork

criticism

  • variety of methods to be used for tertiary structure prediction: I know three types: ab-initio, homology modeling, threading. What else is out there?
    • [2] mentions 4.9 mio hits so i guess there are a variety of ideas and approaches to deal with this topic. Plus [3] specifies in #3 that its a number of things 3>0 and therefore a variety--Hollizeck 21:50, 3 June 2012 (UTC)
    • a-priori
  • What is the purpose of listing the results of three different search tools and not just, for instance, HHsearch?
    • isnt it surprising to you that you do not find at least a similar result when using different approaches? If i do not have a Reference to define right and wrong, how do i know which of the searchtools helps me? Plus what would you do if HHsearch has no result? And just to explain my train of thought:

why do i use an external program to search the pdb, if the pdb does provide one? But then if those results for the low sequence identity do have a bad e-Value, whyy dont i search with another tool to get a better dataset. BUT then this searchtool does not get any result for the middle identity section. Wouldn't you use a third tool to see if there is a explanation for this big difference in the results?

  • Why do you not use common templates of different sequence identity (high, medium, low) for all three methods (Modeller, SWISS-MODEL, I-TASSER)? How will you compare the methods otherwise?
  • Did you compare 2phm to 1pah? They should look identical if the sequence identity is 95%.
    • isn't that kind of our job in this Task? I mean to find out if they look identical with a high identity? --Hollizeck 21:50, 3 June 2012 (UTC)
  • Figure 3: Is the blue structure actually 1pah? It looks different from figure 1 and the thread seems rather unusual.
  • SWISS-MODEL finds of course 1J8U:A since the sequence identity is 100%
    • what is your point?
    • rather: We mention it, to make clear what happens when you use the default settings to predict an already known structure. It chooses one (of several available) known structure (which one? Well, I'd like to know, you might also ;)) and uses this as template. Boidolj 10:38, 4 June 2012 (UTC)
  • There are still plenty of spelling errors.
    • working on it..
  • In the above 80% dataset we find again our structure from above. Would you write this in a paper?
    • i would not, but this is not a paper but a wiki where I can use links and references to make it more dynamic--Hollizeck 21:50, 3 June 2012 (UTC)
  • You might make your plots more uniform, e.g. by using the same background color.
    • good idea --Hollizeck 21:50, 3 June 2012 (UTC)

positive feedback

suggestions

references

<references/>