New alignment feature in dev prev considered harmful

Case number:845813-992845
Topic:Game: Tools
Opened by:Timo van der Laan
Opened on:Sunday, June 3, 2012 - 14:47
Last modified:Monday, June 11, 2012 - 18:58

I really do not like the new functionality in the alignment tool. When aligning it destroys the secondary structure and puts in a new one. It should at least choosable if you want this.
If you want to use the structure predictions and adjust the alignment from that you allways have to restore the prediction.

(Sun, 06/03/2012 - 14:47  |  7 comments)

Joined: 04/15/2012
Groups: Beta Folders

Didn't think of that. It would be good.

Susume's picture
User offline. Last seen 16 hours 21 min ago. Offline
Joined: 10/02/2011

Agreed, it would be good to have the choice. Prior to this functionality, I could judge what to move in the alignment based partly on the secondary structures I had assigned to the protein before aligning. Now those disappear. If I assign helix to a glutamate or loop to a proline and it shows up in the middle of a sheet, I can immediately see there may be a problem with the alignment. If the alignment happily changes it to sheet for me, I have to click each AA in the alignment tool to see what it has put into sheets and what into helices.

beta_helix's picture
User offline. Last seen 4 days 1 hour ago. Offline
Joined: 05/09/2008
Groups: None

The secondary structure is based on the Template's secondary structure that you are threading from.
This is how is used to be until cutpoints introduced a bug that broke this (which is what you still see in the update group: "main")

I don't know if it's possible to add a checkbox for this (Seth?) but if everyone feels the same way (and prefers the bug!) I can just leave them all as loops (which is what all previous CASP10 alignment puzzles had).

I just seemed to us that if the template is a beta barrel, you'd want your model to end up with sheets in the correct places once you thread.

Joined: 09/21/2011
Groups: Void Crushers

With puzzle 566 the models I have seen so far the alignments are pretty close to the prediction. In the earlier puzzles they were not and needed more work which was helped by the prediction. It is a good feature but must be possible to be turned off.

spmm's picture
User offline. Last seen 8 weeks 2 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

can those of us who play main, have lives and work full time, and are now trying to catch up on yet another new change in devprev during CASP have five minutes to check this out before a decision is made please?

I would like to see the structures as provided for each alignment and then evaluate the alignment based on that. thanks

beta_helix's picture
User offline. Last seen 4 days 1 hour ago. Offline
Joined: 05/09/2008
Groups: None

We are sorry that a bug that was introduced is preferred to the way it always used to work with the Alignment Tool.
That is the exact purpose of devprev, to have you let us know what you think about these "bug fixes" and features, so that they get caught BEFORE they are sent to main.

If you "play main, have lives and work full time", then you should be unaffected by this (since it still works as you want it to in main) thanks to those players that do the beta testing for you in devprev: such as this feedback and

Joined: 11/10/2007
Groups: Window Group
Status: Open » Closed

It's been reverted to the old behavior.


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons