Sorry for the delay in responding. I'm really behind on everything these days.
On Thu, Jul 2, 2020 at 11:35 AM Gilberto Pereira
<gilberto.pereiraportugal.gmail.com> wrote:
> Within the context of my work, i was testing out different clustering
> algorithms implemented in the Amber18 package. When testing the Density
> Peaks (DPEAKS) algorithm (Laio and Rodriguez, 2014), i received a warning:
> "Warning: The dpeaks algorithm is still under development. USE WITH
> CAUTION!" My question is the following: Does Amber20 include an optimized
> version of this algorithm?
Not really. I put this warning there because I have only done limited
testing of the algorithm. It seems to work fine, but it's the least
well-tested clustering algorithm in cpptraj, put there by me for fun
one day, so I feel like I need to make people aware that it's a bit
more rough around the edges than the other algorithms.
>
> Additionally, if i try to run the clustering algorithm while employing the
> gauss keyword, CPPTRAJ fails with segfault. Does anyone know why this
> happens?
Oh boy - the 'gauss' keyword is even less tested than 'dpeaks' - mea
culpa. I'll try to look at the code and figure out what may be going
wrong, but you may want to skip that keyword, or maybe choose a more
well-tested algorithm.
Apologies for all the issues.
-Dan
>
> Below is a copy-paste of the segfault:
>
> "
> PARAMETER FILES (1 total):
> 0: ligand.prmtop, 14 atoms, 1 res, box: None, 1 mol
>
> INPUT TRAJECTORIES (1 total):
> 0: 'vac.nc' is a NetCDF AMBER trajectory with coordinates, time, Parm
> ligand.prmtop (reading 2000 of 2000)
> Coordinate processing will occur on 2000 frames.
>
> BEGIN TRAJECTORY PROCESSING:
> .....................................................
> ACTION SETUP FOR PARM 'ligand.prmtop' (3 actions):
> 0: [autoimage]
> Warning: Topology ligand.prmtop does not contain box information.
> Warning: Setup incomplete for [autoimage]: Skipping
> 1: [rms mass first]
> Target mask: [*](14)
> Reference mask: [*](14)
> 2: [createcrd _DEFAULTCRD_]
> Warning: COORDS data sets do not store times.
> Estimated memory usage (2000 frames): 336.024 kB
> ----- vac.nc (1-2000, 1) -----
> 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Complete.
>
> Read 2000 frames and processed 2000 frames.
> TIME: Avg. throughput= 573065.9026 frames / second.
>
> ACTION OUTPUT:
>
> ANALYSIS: Performing 1 analyses:
> 0: [cluster cluster dpeaks epsilon 0.2 dvdfile disvsdens.dat choosepoints
> manual distancecut 0.2 densitycut 0.2 runavg runningave.dat deltafile
> dis-minus-runavg.dat gauss rms mass sieve 10 random out cnumbervtime.dat
> sil Sil summary summary.dat info info.dat cpopvtime cpopvtime.agr normframe
> repout rep repfmt pdb singlerepout represents.nc singlerepfmt netcdf avgout
> AvgFrame_Cluster avgfmt restart]
> Starting clustering.
> Mask [*] corresponds to 14 atoms.
> Random_Number: seed is <= 0, using wallclock time as seed (6365)
> Estimated pair-wise matrix memory usage: > 79.664 kB
> Pair-wise matrix set up with sieve, 2000 frames, 200 sieved frames.
> Calculating pair-wise distances.
> 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Complete.
> Memory used by pair-wise matrix and other cluster data: 88.512 kB
> Starting DPeaks clustering. Using Gaussian kernel to calculate density.
> idx= 398, bandwidth= 0.147002
> Max dist= 1.16382
> Point 7 (frame 208, density 7.31656) selected as candidate for cluster 0
> Point 14 (frame 942, density 6.61173) selected as candidate for cluster 1
> Point 29 (frame 664, density 5.97075) selected as candidate for cluster 2
> Point 60 (frame 1274, density 4.57396) selected as candidate for cluster 3
> Point 68 (frame 113, density 4.40185) selected as candidate for cluster 4
> Point 77 (frame 1737, density 4.09013) selected as candidate for cluster 5
> Point 90 (frame 1566, density 3.33413) selected as candidate for cluster 6
> Point 120 (frame 192, density 2.20433) selected as candidate for cluster 7
> Point 121 (frame 1278, density 2.17273) selected as candidate for cluster 8
> Point 124 (frame 1347, density 2.12788) selected as candidate for cluster 9
> Point 136 (frame 622, density 1.71068) selected as candidate for cluster 10
> Point 149 (frame 643, density 1.42446) selected as candidate for cluster 11
> Point 150 (frame 719, density 1.42032) selected as candidate for cluster 12
> Point 151 (frame 572, density 1.36985) selected as candidate for cluster 13
> Point 158 (frame 1404, density 1.10734) selected as candidate for cluster 14
> Point 160 (frame 495, density 1.0566) selected as candidate for cluster 15
> Point 164 (frame 1157, density 0.981053) selected as candidate for cluster
> 16
> Point 165 (frame 1675, density 0.980377) selected as candidate for cluster
> 17
> Point 169 (frame 371, density 0.924646) selected as candidate for cluster 18
> Point 170 (frame 984, density 0.908064) selected as candidate for cluster 19
> Point 171 (frame 1065, density 0.898682) selected as candidate for cluster
> 20
> Point 173 (frame 1757, density 0.877534) selected as candidate for cluster
> 21
> Point 175 (frame 1544, density 0.840694) selected as candidate for cluster
> 22
> Point 178 (frame 245, density 0.806564) selected as candidate for cluster 23
> Point 180 (frame 481, density 0.781851) selected as candidate for cluster 24
> Point 183 (frame 1038, density 0.630666) selected as candidate for cluster
> 25
> Point 186 (frame 504, density 0.590398) selected as candidate for cluster 26
> Point 187 (frame 1874, density 0.577614) selected as candidate for cluster
> 27
> Point 192 (frame 644, density 0.399897) selected as candidate for cluster 28
> Point 195 (frame 1463, density 0.343491) selected as candidate for cluster
> 29
> Identified 30 cluster centers from density vs distance peaks.
> Killed
> "
>
> Here is the input file i supplied:
>
> "
> parm ligand.prmtop
> trajin vac.nc
> autoimage
> rms mass first
> cluster cluster dpeaks epsilon 0.2 dvdfile disvsdens.dat choosepoints
> manual distancecut 0.2 densitycut 0.2 runavg runningave.dat deltafile
> dis-minus-runavg.dat gauss \
> rms mass \
> sieve 10 random \
> out cnumbervtime.dat \
> sil Sil \
> summary summary.dat \
> info info.dat \
> cpopvtime cpopvtime.agr normframe \
> repout rep repfmt pdb \
> singlerepout represents.nc singlerepfmt netcdf \
> avgout AvgFrame_Cluster avgfmt restart
> run
> go
> quit
> "
>
> Thank you in advance,
> Gilberto Pereira
>
> PhD student at the LABORATOIRE D’INGÉNIERIE DES FONCTIONS MOLÉCULAIRES,
> Univeristy of Strasbourg, France.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Aug 04 2020 - 08:30:02 PDT
