Re: [AMBER] dihedral PCA amber

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Tue, 27 Sep 2016 08:42:11 -0400

Hi,

On Tue, Sep 27, 2016 at 12:09 AM, Neha Gandhi <n.gandhiau.gmail.com> wrote:
> I tried according to your suggestion (Run1) using serial version of
> cpptraj. The program runs without any errors (see blow)

You actually do get some warnings about corrupt frames, e.g.:

> Warning: Frame 37873 coords 1 & 2 overlap at origin; may be corrupt.

You should visualize these frames to make sure everything is OK. If
they are indeed corrupt you should exclude them from this calculation
(via 'check skipbadframes <mask>', where <mask> includes all atoms you
care about).

> but the output
> files dihcovar.dat and modes.dihcovar.dat has NaN NaN..

This could be related to your corrupted frames.

-Dan

>
> I used the same trajectory in CARMA program to calculate dPCA and it works.
>
> Regards,
> Neha
>
>
>
> CPPTRAJ: Trajectory Analysis. V16.00
> ___ ___ ___ ___
> | \/ | \/ | \/ |
> _|_/\_|_/\_|_/\_|_
>
> | Date/time: 09/27/16 12:28:48
> | Available memory: 783.544 MB
>
> Reading 'input.prmtop' as Amber Topology
> INPUT: Reading input from 'ptraj.in'
> [trajin nobadframes.nc]
> Reading 'nobadframes.nc' as Amber NetCDF
> Warning: Trajectory box type is 'Orthogonal' but topology box type is
> 'None'.
> Warning: Setting topology box information from trajectory.
> [multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat]
> MULTIDIHEDRAL: Calculating phi psi dihedrals for residues in range 6-11
> DataSet name: BB
> Output to PhiPsi_6-11.dat
> Output range is -180 to 180 degrees.
> [run]
> ---------- RUN BEGIN -------------------------------------------------
>
> PARAMETER FILES (1 total):
> 0: input.prmtop, 338 atoms, 25 res, box: Orthogonal, 1 mol
>
> INPUT TRAJECTORIES (1 total):
> 0: 'nobadframes.nc' is a NetCDF AMBER trajectory, Parm input.prmtop
> (Orthogonal box) (reading 200000 of 200000)
> Coordinate processing will occur on 200000 frames.
>
> BEGIN TRAJECTORY PROCESSING:
> .....................................................
> ACTION SETUP FOR PARM 'input.prmtop' (1 actions):
> 0: [multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat]
> ResRange=[6-11] phi psi, 12 dihedrals.
> ----- nobadframes.nc (1-200000, 1) -----
> 0% 10% Warning: Frame 37873 coords 1 & 2 overlap at origin; may be corrupt.
> Warning: Frame 37878 coords 1 & 2 overlap at origin; may be corrupt.
> 20% 30% 40% 50% 60% 70% 80% 90% 100% Complete.
>
> Read 200000 frames and processed 200000 frames.
> TIME: Avg. throughput= 14959.3442 frames / second.
>
> ACTION OUTPUT:
>
> DATASETS (12 total):
> BB[phi]:6 "phi:6" (double, torsion(phi)), size is 200000
> BB[psi]:6 "psi:6" (double, torsion(psi)), size is 200000
> BB[phi]:7 "phi:7" (double, torsion(phi)), size is 200000
> BB[psi]:7 "psi:7" (double, torsion(psi)), size is 200000
> BB[phi]:8 "phi:8" (double, torsion(phi)), size is 200000
> BB[psi]:8 "psi:8" (double, torsion(psi)), size is 200000
> BB[phi]:9 "phi:9" (double, torsion(phi)), size is 200000
> BB[psi]:9 "psi:9" (double, torsion(psi)), size is 200000
> BB[phi]:10 "phi:10" (double, torsion(phi)), size is 200000
> BB[psi]:10 "psi:10" (double, torsion(psi)), size is 200000
> BB[phi]:11 "phi:11" (double, torsion(phi)), size is 200000
> BB[psi]:11 "psi:11" (double, torsion(psi)), size is 200000
>
> DATAFILES (1 total):
> PhiPsi_6-11.dat (Standard Data File): phi:6 psi:6 phi:7 psi:7 ... phi:10
> psi:10 phi:11 psi:11
>
> RUN TIMING:
> TIME: Init : 0.0000 s ( 0.00%)
> TIME: Trajectory Process : 13.3696 s ( 82.65%)
> TIME: Action Post : 0.0000 s ( 0.00%)
> TIME: Analysis : 0.0000 s ( 0.00%)
> TIME: Data File Write : 2.8071 s ( 17.35%)
> TIME: Other : 0.0000 s ( 0.00%)
> TIME: Run Total 16.1768 s
> ---------- RUN END ---------------------------------------------------
> [matrix dihcovar dihedrals BB[*] name DIH out dihcovar.dat]
> MATRIX: Calculating dihedral covariance matrix, output is by atom.
> 12 data sets.
> Printing to file dihcovar.dat
> Matrix data set is 'DIH'
> Start: 1 Stop: Final frame
> [diagmatrix DIH vecs 4 out modes.dihcovar.dat name DIHMODES]
> Changed DataFile 'modes.dihcovar.dat' type to Evecs file for set
> DIHMODES
> DIAGMATRIX: Diagonalizing matrix DIH and writing modes to
> modes.dihcovar.dat
> Calculating 4 eigenvectors.
> Storing modes with name: DIHMODES
> [run]
> ---------- RUN BEGIN -------------------------------------------------
>
> PARAMETER FILES (1 total):
> 0: input.prmtop, 338 atoms, 25 res, box: Orthogonal, 1 mol
>
> INPUT TRAJECTORIES (1 total):
> 0: 'nobadframes.nc' is a NetCDF AMBER trajectory, Parm input.prmtop
> (Orthogonal box) (reading 200000 of 200000)
> Coordinate processing will occur on 200000 frames.
>
> BEGIN TRAJECTORY PROCESSING:
> .....................................................
> ACTION SETUP FOR PARM 'input.prmtop' (1 actions):
> 0: [matrix dihcovar dihedrals BB[*] name DIH out dihcovar.dat]
> ----- nobadframes.nc (1-200000, 1) -----
> 0% 10% Warning: Frame 37873 coords 1 & 2 overlap at origin; may be corrupt.
> Warning: Frame 37878 coords 1 & 2 overlap at origin; may be corrupt.
> 20% 30% 40% 50% 60% 70% 80% 90% 100% Complete.
>
> Read 200000 frames and processed 200000 frames.
> TIME: Avg. throughput= 88342.5394 frames / second.
>
> ACTION OUTPUT:
>
>
>
>
>
>
>
>
>
>
>
>
> On 23 September 2016 at 22:11, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>
>> Ah, I see the issue now - you're running the MPI version of cpptraj
>> and trying to do 2 trajectory processing runs in the same execution
>> which is currently not supported (and something cpptraj should check
>> for - details below for those interested). You can still perform the
>> analysis, you just need to separate it into two runs; one where you
>> generate the eigenvectors and the second where you project, e.g.
>>
>> # Run 1
>> trajin x.mdcrd
>> multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat
>> run
>> # Calculate dihedral covariance matrix and obtain eigenvectors
>> matrix dihcovar dihedrals BB[*] out dihcovar.dat name DIH
>> diagmatrix DIH vecs 4 out modes.dihcovar.dat name DIHMODES
>> run
>>
>>
>> # Run 2
>> trajin x.mdcrd
>> # Read in previously generated dihedral data
>> readdata PhiPsi_6-11.dat name BB
>> # Make sure cpptraj knows its dihedral data
>> dataset mode torsion BB
>> # Read in previously generated eigenvectors
>> readdata modes.dihcovar.dat name MyModes
>> # Project along eigenvectors
>> projection evecs modes MyModes out dih.project.dat beg 1 end 4 dihedrals BB
>> run
>>
>> This way should work. Thanks for the report.
>>
>> -Dan
>>
>> On Thu, Sep 22, 2016 at 10:00 PM, Neha Gandhi <n.gandhiau.gmail.com>
>> wrote:
>> > Hi Daniel,
>> >
>> > I applied recent patches to AmberTools 16. Here is the output
>> > CPPTRAJ: Trajectory Analysis. V16.12 MPI
>> > ___ ___ ___ ___
>> > | \/ | \/ | \/ |
>> > _|_/\_|_/\_|_/\_|_
>> >
>> > | Running on 6 threads
>> > | Date/time: 09/23/16 11:53:36
>> > | Available memory: 21.905 GB
>> >
>> > Reading 'input.prmtop' as Amber Topology
>> > INPUT: Reading input from 'ptraj.in'
>> > [trajin nobadframes.nc]
>> > Reading 'nobadframes.nc' as Amber NetCDF
>> > Warning: Trajectory box type is 'Orthogonal' but topology box type is
>> > 'None'.
>> > Warning: Setting topology box information from trajectory.
>> > [multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat]
>> > MULTIDIHEDRAL: Calculating phi psi dihedrals for residues in range
>> 6-11
>> > DataSet name: BB
>> > Output to PhiPsi_6-11.dat
>> > Output range is -180 to 180 degrees.
>> > [run]
>> > ---------- RUN BEGIN -------------------------------------------------
>> >
>> > PARAMETER FILES (1 total):
>> > 0: input.prmtop, 338 atoms, 25 res, box: Orthogonal, 1 mol
>> >
>> > INPUT TRAJECTORIES (1 total):
>> > 0: 'nobadframes.nc' is a NetCDF AMBER trajectory, Parm input.prmtop
>> > (Orthogonal box) (reading 200000 of 200000)
>> > Coordinate processing will occur on 200000 frames.
>> >
>> > PARALLEL INFO:
>> > Thread 0 will process 33334 frames.
>> > Thread 1 will process 33334 frames.
>> > Thread 2 will process 33333 frames.
>> > Thread 3 will process 33333 frames.
>> > Thread 4 will process 33333 frames.
>> > Thread 5 will process 33333 frames.
>> > .....................................................
>> > ACTION SETUP FOR PARM 'input.prmtop' (1 actions):
>> > 0: [multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat]
>> > ResRange=[6-11] phi psi, 12 dihedrals.
>> >
>> > BEGIN PARALLEL TRAJECTORY PROCESSING:
>> > 0% 10% [1] Warning: Set 37873 coords 1 & 2 overlap at origin; may be
>> > corrupt.
>> > [1] Warning: Set 37878 coords 1 & 2 overlap at origin; may be corrupt.
>> > 20% 30% 40% 50% 60% 70% 80% 90% 100% Complete.
>> > TIME: Rank 0 throughput= 204.0497 frames / second.
>> > TIME: Rank 1 throughput= 203.6889 frames / second.
>> > TIME: Rank 2 throughput= 204.1462 frames / second.
>> > TIME: Rank 3 throughput= 203.3635 frames / second.
>> > TIME: Rank 4 throughput= 203.6441 frames / second.
>> > TIME: Rank 5 throughput= 203.7433 frames / second.
>> > TIME: Avg. throughput= 1220.1928 frames / second.
>> >
>> > ACTION OUTPUT:
>> >
>> > DATASETS (12 total):
>> > BB[phi]:6 "phi:6" (double, torsion(phi)), size is 200000
>> > BB[psi]:6 "psi:6" (double, torsion(psi)), size is 200000
>> > BB[phi]:7 "phi:7" (double, torsion(phi)), size is 200000
>> > BB[psi]:7 "psi:7" (double, torsion(psi)), size is 200000
>> > BB[phi]:8 "phi:8" (double, torsion(phi)), size is 200000
>> > BB[psi]:8 "psi:8" (double, torsion(psi)), size is 200000
>> > BB[phi]:9 "phi:9" (double, torsion(phi)), size is 200000
>> > BB[psi]:9 "psi:9" (double, torsion(psi)), size is 200000
>> > BB[phi]:10 "phi:10" (double, torsion(phi)), size is 200000
>> > BB[psi]:10 "psi:10" (double, torsion(psi)), size is 200000
>> > BB[phi]:11 "phi:11" (double, torsion(phi)), size is 200000
>> > BB[psi]:11 "psi:11" (double, torsion(psi)), size is 200000
>> >
>> > DATAFILES (1 total):
>> > PhiPsi_6-11.dat (Standard Data File): phi:6 psi:6 phi:7 psi:7 ...
>> phi:10
>> > psi:10 phi:11 psi:11
>> >
>> > RUN TIMING:
>> > TIME: Init : 0.0004 s ( 0.00%)
>> > TIME: Trajectory Process : 163.9085 s ( 80.98%)
>> > TIME: Data Set Sync : 0.0093 s ( 0.00%)
>> > TIME: Action Post : 0.0001 s ( 0.00%)
>> > TIME: Analysis : 0.0000 s ( 0.00%)
>> > TIME: Data File Write : 38.4950 s ( 19.02%)
>> > TIME: Other : 0.0002 s ( 0.00%)
>> > TIME: Run Total 202.4135 s
>> > ---------- RUN END ---------------------------------------------------
>> > [matrix dihcovar dihedrals BB[*] name DIH out dihcovar.dat]
>> > MATRIX: Calculating dihedral covariance matrix, output is by atom.
>> > 12 data sets.
>> > Printing to file dihcovar.dat
>> > Matrix data set is 'DIH'
>> > Start: 1 Stop: Final frame
>> > [runanalysis diagmatrix DIH vecs 2 out modes.dihcovar.dat name
>> DIHMODES]
>> > Changed DataFile 'modes.dihcovar.dat' type to Evecs file for set
>> > DIHMODES
>> > DIAGMATRIX: Diagonalizing matrix DIH and writing modes to
>> > modes.dihcovar.dat
>> > Calculating 2 eigenvectors.
>> > Storing modes with name: DIHMODES
>> > Eigenmode calculation for 'DIH'
>> > Warning: Specified # of eigenmodes to calc (2) > matrix dimension (0).
>> > Warning: Only calculating 0 eigenmodes.
>> > Calculating eigenvectors and eigenvalues.
>> > Calculating all eigenmodes.
>> > ** On entry to DSPEV parameter number 7 had an illegal value
>> > ** On entry to DSPEV parameter number 7 had an illegal value
>> > ** On entry to DSPEV parameter number 7 had an illegal value
>> > ** On entry to DSPEV parameter number 7 had an illegal value
>> > ** On entry to DSPEV parameter number 7 had an illegal value
>> >
>> >
>> >
>> >
>> > On 23 September 2016 at 05:16, Daniel Roe <daniel.r.roe.gmail.com>
>> wrote:
>> >
>> >> Can you post the exact version of cpptraj you are using as well as all
>> >> output? Thanks.
>> >>
>> >> -Dan
>> >>
>> >> On Wed, Sep 21, 2016 at 11:08 PM, Neha Gandhi <n.gandhiau.gmail.com>
>> >> wrote:
>> >> > I am trying to do dihedral PCA and I am following the example in
>> AMBER16
>> >> > manual
>> >> >
>> >> > trajin x.mdcrd
>> >> > multidihedral BB phi psi resrange 6-11 out PhiPsi_6-11.dat
>> >> > run
>> >> > # Calculate dihedral covariance matrix and obtain eigenvectors
>> >> > matrix dihcovar dihedrals BB[*] out dihcovar.dat name DIH
>> >> > diagmatrix DIH vecs 4 out modes.dihcovar.dat name DIHMODES
>> >> > run
>> >> > # Project along eigenvectors
>> >> > projection evecs DIHMODES out dih.project.dat beg 1 end 4 dihedrals
>> BB[*]
>> >> > run
>> >> >
>> >> >
>> >> >
>> >> > I get error-It is not able to read BB[*] data set. I can see that the
>> >> > output file PhiPsi_6-11.dat is not empty.
>> >> >
>> >> > Any help is appreciated.
>> >> >
>> >> >
>> >> > --
>> >> > Regards,
>> >> > Dr. Neha S. Gandhi,
>> >> > Vice Chancellor's Research Fellow,
>> >> > Queensland University of Technology,
>> >> > 2 George Street, Brisbane, QLD 4000
>> >> > Australia
>> >> > LinkedIn
>> >> > Research Gate
>> >> > _______________________________________________
>> >> > AMBER mailing list
>> >> > AMBER.ambermd.org
>> >> > http://lists.ambermd.org/mailman/listinfo/amber
>> >>
>> >>
>> >>
>> >> --
>> >> -------------------------
>> >> Daniel R. Roe
>> >> Laboratory of Computational Biology
>> >> National Institutes of Health, NHLBI
>> >> 5635 Fishers Ln, Rm T900
>> >> Rockville MD, 20852
>> >> https://www.lobos.nih.gov/lcb
>> >>
>> >> _______________________________________________
>> >> AMBER mailing list
>> >> AMBER.ambermd.org
>> >> http://lists.ambermd.org/mailman/listinfo/amber
>> >>
>> >
>> >
>> >
>> > --
>> > Regards,
>> > Dr. Neha S. Gandhi,
>> > Vice Chancellor's Research Fellow,
>> > Queensland University of Technology,
>> > 2 George Street, Brisbane, QLD 4000
>> > Australia
>> > LinkedIn
>> > Research Gate
>> > _______________________________________________
>> > AMBER mailing list
>> > AMBER.ambermd.org
>> > http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>>
>> --
>> -------------------------
>> Daniel R. Roe
>> Laboratory of Computational Biology
>> National Institutes of Health, NHLBI
>> 5635 Fishers Ln, Rm T900
>> Rockville MD, 20852
>> https://www.lobos.nih.gov/lcb
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
>
>
> --
> Regards,
> Dr. Neha S. Gandhi,
> Vice Chancellor's Research Fellow,
> Queensland University of Technology,
> 2 George Street, Brisbane, QLD 4000
> Australia
> LinkedIn
> Research Gate
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber



-- 
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Sep 27 2016 - 06:00:03 PDT
Custom Search