You are here

How to get the GDT_TS?

4 posts / 0 new
Last post
How to get the GDT_TS?

Dear all,

I have got results (silent and score files)for many targets. Although, I provided the native file, I couldn't find the GDT_TS in the score file.

I have got the rmsd, but I need both rmsd and GDT_TS. Is there any Rosetta's tool (or even an independent tool) to calculate the GDT_TS of a native structure against a silent file?

Thank you in advance,

Post Situation: 
Thu, 2013-09-05 11:59

Rosetta tends to calls GDT_TS "GDT_MM". (From what I understand, the MM stands for MAMMOTH, an alignment technique, as different alignment techniques give subtly different GDT results.)

The "score" application (although not "score_jd2") should give you the GDT in the output scorefile if you provide the native structure with the -in:file:native flag (note that the native has to be the same length as all the structures being analyzed). There's a number of columns labeled gdtmm_X_Y for the sub components, where X is the distance tolerance, and Y is the rms tolerance. The plain gdtmm is the average of gdtmm_1_1, gdtmm_2_2, gdtmm_3_3, gdtmm_4_3, and gdtmm_7_4

Before running the score application, I'd check your existing scorefiles. Many of the ab initio/comparative modeling applications will also output the GDT metrics if you have provided them with the native PDB.

Thu, 2013-09-05 15:02

Hi again,

I ran the "score" application on my silent file providng the native structure and I got a file called However, it is not quite different from the file "score.fsc" I had already had except for the following new terms:

- irms
- srms
- silent_score

Question 1: what are these?
Question 2: No GDT scores were shown

I would appreciate if you could help me.


Thu, 2013-09-12 04:31

"irms" is the C alpha rmsd of the final pose to the input pose (should be zero for the scoring application, but will be nonzero for some other applications)

"rms" is the C alpha rmsd of the final pose to the "native" pose (the one passed to -in:file:native)

"srms" is the C alpha rmsd of the *input* pose to the native pose

"silent_score" would be the score copied from the input silent file

Regarding the missing GDT scores, one possibility is that your native pose doesn't have the same number of residues as the final pose. There's an explicit check for this case, and won't calculate the GDT statistics if there's a length mismatch. (There's no such check for the above rms statistics - it will blindly match up one-to-one, with any extra bit hanging off the end.)

If you think the two structures should have the same number of residues, one thing to check is the occupancy column of the PDB. If it's zero, Rosetta will discard those atoms. If certain atoms (basically the backbone atoms) are missing (either from being discarded or not existing in the PDB in the first place) Rosetta will discard the residue completely. It typically prints a message to that effect in the log, though.

Thu, 2013-09-12 10:51