You are here

problems in InterfaceAnalyzer analysis

10 posts / 0 new
Last post
problems in InterfaceAnalyzer analysis
#1

Hi, everyone!

Now I met a samll bug during InterfaceAnalyzer analysis as follows.

`core.io.silent: Reading all structures from 1-46wt.out
core.io.silent: bad format in sequence line of silent file
core.io.silent: line = SEQUENCE:
core.io.silent: tag = SEQUENCE:
terminate called after throwing an instance of 'utility::excn::EXCN_BadInput'

And the starting lines of my silent file are:
`SEQUENCE:
SCORE: total_score Binding energy dslf_ca_dih dslf_cs_ang dslf_ss_dih dslf_ss_dst fa_atr fa_dun fa_intra_rep fa_pair fa_rep fa_sol hack_elec hbond_bb_sc hbond_lr_bb hbond_sc hbond_sr_bb omega p_aa_pp pro_close rama ref description`

In addition, I also tried to delete` SEQUENCE:` line

And this time, error message was:
`
core.io.silent: bad format in first line of silent file 1-46wt.out (function read_stream):

core.io.silent: SCORE: total_score Binding_energy dslf_ca_dih dslf_cs_ang dslf_ss_dih dslf_ss_dst fa_atr fa_dun fa_intra_rep fa_pair fa_rep fa_sol hack_elec hbond_bb_sc hbond_lr_bb hbond_sc hbond_sr_bb omega p_aa_pp pro_close rama ref description

terminate called after throwing an instance of 'utility::excn::EXCN_BadInput'

`

Waiting for any suggestion!
Thanks very much.

Post Situation: 
Fri, 2011-12-02 23:36
chenchaozhao

I think you need a flag for telling Rosetta which silent file type you have. I don't know what it is, unfortunately. Look for something like in:file:silent_struct_type, perhaps? If you can't find it I'll look harder later.

Sat, 2011-12-03 06:21
smlewis

Hi, Steven:

Thanks for your advice!

I have tried to add -in:file:silent_struct_type protein flag either to options or to commands. Both of them showed "-in:file:silent_struct_type" was not a regular flag in command lines.

And then I checked my silent file as follows.

"SCORE: total_score Binding energy dslf_ca_dih dslf_cs_ang dslf_ss_dih dslf_ss_dst fa_atr fa_dun fa_intra_rep fa_pair fa_rep fa_sol hack_elec hbond_bb_sc hbond_lr_bb hbond_sc hbond_sr_bb omega p_aa_pp pro_close rama ref description
SCORE: 1824.405 -4.117 0.000 0.000 0.000 0.000 -1081.145 505.275 3.512 -29.290 2082.608 586.912 -19.670 -12.650 -6.378 -7.809 -100.221 5.974 -21.167 7.964 15.101 -104.610 wt_0001
SCORE: 1931.015 3.728 0.000 0.000 0.000 0.000 -1074.671 516.308 3.545 -26.844 2171.753 579.399 -18.603 -14.052 -7.861 -4.482 -97.903 5.641 -20.140 7.899 15.635 -104.610 wt_0002

...."

So I suddenly realized that this silent file was exactly about enery calculation from floppytail application, different from other silent files which can be used to extract structures,such as in abinitio modeling.

So I had to use -in:file:s or -in:file:l for input structures, but not -in:file:silent.

Am I right?

Thanks for your clarification in advance.

Sat, 2011-12-03 22:03
chenchaozhao

A) The flag is "in:file:silent_struct_type"

B) That looks like it's just scores, no coordinates, so InterfaceAnalyzer won't work; it needs structures. There is a silent file type for just scores; there's also the functionally identical "scorefile" that you get when using PDB output. It's not clear to me if what you've pasted is from a score-only silent file, or a scorefile.

C) Yes, you should use -s/-l if you have PDBs. InterfaceAnalyzer will work with silent files that have coordinates but yours appears not to.

Mon, 2011-12-05 08:15
smlewis

Hi, everyone.

I have another problem when scaling structures with dGseparate.

My question is whether dGseparate (binding energy) should be evaluated identical to total enery, I mean, the lower value means the better compacted complex, or on the opposite.

I found all the dGseparate values were positive, if binding energy= AB-A-B, I think all of them should have been minus. Whether I should still select the structures with lower value of dGseparate in this case?

Thanks.

Mon, 2012-01-02 04:13
chenchaozhao

dGseparate is an energy: negative is good, positive is bad.

dGseparate is unlike total_score in that it usually won't be "poisoned" into being a positive value by constant, irrelevant clashes. The subtraction step controls for that. If you are getting positive binding energies it should be concerning and you ought to determine why instead of just picking the best and moving on. (That said, if there is a clash in an _unmodeled_ part of the interface, it WILL give you behaviors like this, so perhaps you can just pick the best score and move on).

Mon, 2012-01-02 11:06
smlewis

Thanks for your explanation.

My following question is when I checked the binding energy of these structures in Floppytail, they showed negative values. However, when I reevaluated them, same structures, by InterfaceAnalyzer, both dGseparate and dGgross of them were positive values. Is it possible, and which should be given priority for evaluation?

Also, whether generally low total score indicated better structure than that with high total score? Sorry, I was a bit confused.

Mon, 2012-01-02 18:40
chenchaozhao

I don't understand what you mean in the first part of your post - I'm pretty sure FloppyTail doesn't do dGbind and dGcross, and if it does it's via the InterfaceAnalzyerMover. What do you mean by "checked the binding energy in FloppyTail"?

For the second part - low total scores are better.

Tue, 2012-01-03 17:38
smlewis

Hi, thanks for your response.

In first part of my last post, I meant during FloppyTail analysis, I got binding energy calculation results, like follows.

SCORE: total_score Binding energy dslf_ca_dih dslf_cs_ang dslf_ss_dih dslf_ss_dst fa_atr fa_dun fa_intra_rep fa_pair fa_rep fa_sol hack_elec hbond_bb_sc hbond_lr_bb hbond_sc hbond_sr_bb omega p_aa_pp pro_close rama ref description
SCORE: 1373.242 -9.4 0 0 0 0 -1066.803 544.994 6.63 -28.945 1575.833 560.357 -13.772 -14.121 -6.378 -5.931 -98.509 17.264 -20.819 7.841 20.213 -104.61 wt_2454
SCORE: 1793.338 -2.172 0 0 0 0 -1087.07 582.839 4.467 -24.454 1982.16 578.028 -20.568 -12.904 -6.378 -3.991 -99.011 6.767 -22.736 6.782 14.018 -104.61 wt_1829
SCORE: 2255.083 -10.432 0 0 0 0 -1070.135 532.831 3.641 -27.74 2484.228 569.975 -15.102 -13.701 -6.378 -4.187 -100.396 6.869 -21.268 6.725 14.331 -104.61 wt_1080
SCORE: 2288.325 0.011 0 0 0 0 -1100.879 580.528 3.95 -25.975 2486.531 585.956 -20.368 -12.713 -6.378 -4.129 -101.435 9.629 -23.577 7.174 14.62 -104.61 wt_1429
SCORE: 2353.817 -1.017 0 0 0 0 -1092.82 535.082 3.807 -24.124 2578.016 586.225 -20.945 -11.99 -6.528 -3.314 -98.28 10.726 -21.951 8.335 16.188 -104.61 wt_1803
.....

Such binding energy calculation was really following your instruction, to activate binding energy calculation which was orginally turned off in floppytail. You helped me to solve that problem in
the post "a trouble during floppytail modeling", especially your response in #15 and #17.

From the data I listed above, you could find that the binding energy were negative. But when I reevaluated the same structures in InterfaceAnalyzer, dGsepa and dGgross, standing for binding energy, were totally positive.

My question is which result is more reliable?

Thanks.

Tue, 2012-01-03 17:54
chenchaozhao

I would guess that FloppyTail and InterfaceAnalyzer are calculating on different interfaces. I re-read the old forum thread in which I advised you to tweak the "create_extra_output" function. I would guess that your rewriting of that function is splitting the pose up differently than InterfaceAnalyzer does. Assuming you wrote the binding energy calculation in your tweak of FloppyTail correctly, I would trust it more; InterfaceAnalyzer may be calculating the wrong interface. I'm not 100% sure of this.

Wed, 2012-01-04 07:42
smlewis