error_ ensemble NMA

[karolk]

Hi,

I am trying to perform ensemble NMA.

I am getting weird error: "Warning in mylapply(1:length(pdbs$id), .calcAlnModes, pdbs, xyz, gaps.res, : all scheduled cores encountered errors in user code Error in tmp.modes$U : $ operator is invalid for atomic vectors"

code:

pdbs.list <-list.files(path="path",pattern="^pdb_a_")
align.pdbs <- pdbaln(pdbs.list,fit = FALSE)
modes <- nma(align.pdbs,keep = 50, mass = TRUE, ff="calpha", ncore=2, fit=FALSE)

thank you

[Xinqiu Yao]

Could you provide an example pdbs object that we can reproduce your problems? You can either attach the file here or send it to [e-mail].

[karolk]

Hi Yao,

The problem was ( I think) that I was operating on Calphas only. When I pdbaln whole structures, the problem disappeared. Nevertheless, I encountered another problem.

Error: ... dimension of x$U.subspace: ( 13944x13938x4 ) ... aligned eigenvectors (gap containing positions removed)
... estimated memory usage of final 'eNMA' object: 5931.6 Mb

| | 0%Error in modes$U[, k] : subscript out of bounds

I managed to reproduce the problem. The argument, which is causing the problem is: "keep=30". Without it, eNMA goes fine. Nevertheless, I would prefer to store only first 30.. If I store all modes for my protein, the binaries go really big and working with these is quite inconvenient.

The problem appears on two different platforms:

(1) Mac Os, R version 3.4.1 (2017-06-30)

(2) Linux cluster, R version 3.4.2 (2017-09-28)

The code is exactly the same.

pdbs.list <-list.files(path="path",pattern="^pdb_a_")
align.pdbs <- pdbaln(pdbs.list,fit = FALSE)
modes <- nma(align.pdbs,keep = 30, mass = TRUE, ff="calpha", ncore=2, fit=FALSE)

[Xinqiu Yao]

Hi,

Instead of using keep=30, try subspace=30 (See help(nma.pdbs) for more detail).

(To Lars: We need to override 'subspace' when users set 'keep' in calling nma.pdbs() (otherwise, 'out of boundary' errors). Or, we can modify to ignore 'keep' in nma.pdbs(). What do you think?)

[karolk]

Thanks Yao. I will try and I will let you know. cheers

[karolk]

Worked. Please, close this ticket.

[Lars Skjærven]

(To Lars: We need to override 'subspace' when users set 'keep' in calling nma.pdbs() (otherwise, 'out of boundary' errors). Or, we can modify to ignore 'keep' in nma.pdbs(). What do you think?)