error using deconvolution_nmf

[Sames-Jtudd]

Hi, Im having some issues with the above command. Hope you can help. Any ideas you had would be very useful.

When running the following command

denovo_signatures <- deconvolution_nmf(input_data = propMutsByCat, type = "SNV", range_of_sigs = 2:10 ,nrun = 20 ,method = "brunet" ,resdir = resdir)
I get the following output

[1] "Estimating the optimal number of mutational signatures..."
Timing stopped at: 1.475 2.602 4.144
Timing stopped at: 1.68 2.829 4.582
Timing stopped at: 1.478 2.621 4.167
Timing stopped at: 1.651 2.82 4.545
Timing stopped at: 1.477 2.622 4.166
Timing stopped at: 1.666 2.826 4.565
Timing stopped at: 1.497 2.586 4.206
Timing stopped at: 1.656 2.74 4.471
Timing stopped at: 1.478 2.668 4.21
Error in (function (...) : All the runs produced an error:
-#1 [r=2] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 2, 0), method = "random")
  -Cancer cell fraction > 1.0 #2 [r=3] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 3, 0), method = "random")
  -error when preprocessInput_snv #3 [r=4] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 4, 0), method = "random")
  -What is "ntot" #4 [r=5] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 5, 0), method = "random")
  -SV data preparing #5 [r=6] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 6, 0), method = "random")
  -Error when using preprocessInput_snv with example data #6 [r=7] -> NMF::nmf - 20/20 fit(s) threw an error.

Error(s) thrown:

• run preprocessInput_snv throw errors using example data  #1: unused arguments (model = list("NMFstd", 7, 0), method = "random")
  -Error in annotating the data #7 [r=8] -> NMF::nmf - 20/20 fit(s) threw an

given that the input data is generated using a Palimpsest command Im surprised that this threw an error.

hear is a sample of the input file

head(propMutsByCat)
Sample1 Sample2 Sample3 Sample4 Sample5
CA_A.A 0.025848142 0.017460317 0.019760790 0.017004578 0.0192307692
CA_A.C 0.011308562 0.015873016 0.014040562 0.015042511 0.0142679901
CA_A.G 0.001615509 0.003174603 0.003120125 0.002616089 0.0006203474
CA_A.T 0.016962843 0.012698413 0.010400416 0.012426422 0.0130272953
CA_C.A 0.016962843 0.017460317 0.015600624 0.017004578 0.0136476427
CA_C.C 0.010500808 0.012698413 0.011960478 0.010464356 0.0068238213

thanks
Jamie

[jh2663]

I had very similar kind of errors as shown here. My solutions were to:

1. give a try reinstallation of 'NMF' package with ref='devel' option:

devtools::install_github("renozao/NMF",ref = 'devel', force=TRUE)

2. use different NMF algorithms such as 'lee' or 'nsNMF' instead of 'brunet'

My case had very few data points so 'brunet' algorithm failed to calculate the matrix properly for some reasons. Other algorithms work fine even for very sparse input matrix. FYI, I find 'nsNMF' looked more reasonable to my case.

Hope this helps,

[Sames-Jtudd]

cheers ill give that a go.

[Sames-Jtudd]

So that didn't work sadly.

[FunGeST]

Hi Jamie,

Thanks for getting in touch, and I'm sorry for not getting back to you sooner.

I haven't seen this issue before, even when working with sparse datasets as @jh2663 suggested. I notice you're using the older version of Palimpsest, would you be able to update your version of the package and let us know if you're still having the same problem? Hopefully that should fix it or at least make it easier for us to help you!

The latest version can be installed like this:

install.packages("devtools")
library(devtools)
devtools::install_github("FunGeST/Palimpsest", force = TRUE)

best wishes,
Benedict

[Sames-Jtudd]

Hi Benedict, Thanks for getting back to me, thought I had the most recent version, I’ll try that and fingers crossed it will work. Best Jamie From: FunGeST <[email protected]> Sent: 13 January 2020 11:53 To: FunGeST/Palimpsest <[email protected]> Cc: James Studd <[email protected]>; Author <[email protected]> Subject: Re: [FunGeST/Palimpsest] error using deconvolution_nmf (#34) Hi Jamie, Thanks for getting in touch, and I'm sorry for not getting back to you sooner. I haven't seen this issue before, even when working with sparse datasets as @jh2663<https://github.com/jh2663> suggested. I notice you're using the older version of Palimpsest, would you be able to update your version of the package and let us know if you're still having the same problem? Hopefully that should fix it or at least make it easier for us to help you! The latest version can be installed like this: install.packages("devtools") library(devtools) devtools::install_github("FunGeST/Palimpsest", force = TRUE) best wishes, Benedict — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#34?email_source=notifications&email_token=AICFCUNSTJGGWEEORO6HHDLQ5RIYJA5CNFSM4KCNZGL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIYNZMY#issuecomment-573627571>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AICFCULXXWNBLXX3LD62FELQ5RIYJANCNFSM4KCNZGLQ>. The Institute of Cancer Research: Royal Cancer Hospital, a charitable Company Limited by Guarantee, Registered in England under Company No. 534147 with its Registered Office at 123 Old Brompton Road, London SW7 3RP. This e-mail message is confidential and for use by the addressee only. If the message is received by anyone other than the addressee, please return the message to the sender by replying to it and then delete the message from your computer and network.

[FunGeST]

Thanks! Please note that some of the function names have now changed, see the example script for more information on this (e.g. NMF_Extraction() has replaced deconvolution_nmf()).