[gmx-users] Regarding DPCA

[Matthias Ernst]

Sorry to pop in without actually addressing your questions.
But, as the question for dPCA comes up every year or so, let me just 
state what I wrote some time ago:

https://www.mail-archive.com/gromacs.org_gmx-users@maillist.sys.kth.se/msg24762.html

Tl;dr:
In my opinion, the way dPCA is described on the GROMACS page is rather 
misleading and prone to errors. Don't do it that way.
Better use some tool of your choice to write out the dihedral angles as 
ASCII file (it's much easier to see if everything worked out in this 
case anyway) and use FastPCA (https://github.com/lettis/FastPCA) for 
actually performing the PCA. All data (covariance matrix, eigenvectors 
and -values and principal components) then are generated as ASCII files 
as well and can be analysed and plotted with the tool of your choice.

In addition, please note that it is possible to perform dPCA without 
doing the sine/cosine transformation, see 
https://doi.org/10.1063/1.4998259. We recommend using this procedure as 
in this case, the covariance matrix and eigenvectors can readily be 
interpreted (e.g., dihedral x is the moste important one...)

Best regards,
Matthias


Am 05.02.2018 um 10:17 schrieb Seera Suryanarayana:
>> Dear gromacs users
>>
>> I done simulations of peptide of 20 residues length for 300ns. As I would
>> like to explore the conformational flexibility I have chosen to do dPCA. I
>> have gone through the tutorial which is present in the gromacs cite.
>> Firstly, I have created the index file which shows the dihedral angle atom
>> numbers. I have given the 36 dihedral angles information of 18 residues.
>> Based on the formula 2*N/3 for 24 atoms I can create index file which can
>> be used in the neat step to create the .gro file. In my index file I gave
>> first 24 backbone atoms numbers. When  use the gmx trjconv for getting
>> resized.gro I got following error.
>>
>> Fatal error:
>> Index[4] 25 is larger than the number of atoms in the
>> trajectory file (24). There is a mismatch in the contents
>> of your -f, -s and/or -n files.
>>
>> I got error meaning. In my index file fifth element is 25 which is greater
>> than the total number of atoms number that is 24. If I give first 24
>> numbers Then I got .gro file only for first file as follow.
>>
>> Generated by trjconv : Protein in water t=   0.00000
>>      24
>>       1LYS      N    1  -0.370  -0.929   0.912
>>       1LYS     H1    2   0.411   0.167  -0.986
>>       1LYS     H2    3  -0.533   0.846  -0.352
>>       1LYS     H3    4  -0.936  -0.603   0.798
>>       1LYS     CA    5   0.049  -0.999  -1.000
>>       1LYS     HA    6  -0.003   0.264  -0.965
>>       1LYS     CB    7   0.935  -0.356   0.188
>>       1LYS    HB1    8  -0.982   0.862  -0.506
>>       1LYS    HB2    9   0.358  -0.934   0.839
>>       1LYS     CG   10  -0.544   0.010  -1.000
>>       1LYS    HG1   11  -0.646   0.763  -0.335
>>       1LYS    HG2   12  -0.942  -0.930  -0.367
>>       1LYS     CD   13   0.374  -0.927  -0.662
>>       1LYS    HD1   14   0.750   0.341  -0.940
>>       1LYS    HD2   15  -0.661   0.751  -0.775
>>       1LYS     CE   16  -0.632  -0.591   0.807
>>       1LYS    HE1   17  -0.561   0.828   0.693
>>       1LYS    HE2   18   0.721   0.610  -0.793
>>       1LYS     NZ   19  -0.410   0.912  -0.585
>>       1LYS    HZ1   20  -0.811  -0.419   0.908
>>       1LYS    HZ2   21   0.054  -0.999  -0.159
>>       1LYS    HZ3   22   0.987   0.203  -0.979
>>       1LYS      C   23   0.749  -0.663  -0.820
>>       1LYS      O   24  -0.573  -0.900   0.436
>>      2.00000   2.00000   2.00000
>>
>> But I want to get the information for first 10 residues of my peptide.
>> Kindly tell me how to do get dPCA for first ten residues of my 20 residues
>> simulated peptide.
> 
> It seems you only saved the first residue in your trajectory. If you
> make matching trajectory and .tpr files that save the first ten
> residues, then you should be able to proceed.
> 
> -Justin
> 
> Dear Justin
> 
> In brief I would like to tell you my problem. I have taken the atom numbers
> those represent the dihedral angle atoms from the topology file to generate
> the index file to generate .trr file. Then I have generated the another
> index file for .gro file. As number of dihedral angles 36 for 18 residues,
> I could save 24 atoms according to 2*N/3 formula. As my area of interest is
> first 2-12 residues, I have taken numbers of backbone atoms in my index
> file for resized.gro file. Then I executed the following command
> 
> trjconv -s foo.tpr -f dangle.trr -o resized.gro -n covar.ndx -e 0
> 
> 
> Here I got the error as follow
> 
>> Fatal error:
>> Index[4] 25 is larger than the number of atoms in the trajectory file
> (24). There is a mismatch in the contents of your -f, -s and/or -n files.
> 
> Now come to your suggestion, you suggested me that first make matching
> trajectory and .tpr files for 10 residues. Here I not able to getting how
> does that work. Because we have to generate the .trr file based on the
> dihedral index file, which means definitely I will get the same error.
> Kindly tell me more about how to generate the resized.gro file for first 10
> residues.
> 
> Thanks in advance
> 
> 
> 
> Surya
> Graduate student
> India.
>