segfault: linked to #73

[susheelbhanu]

Hi @zdk123,

Since #73 is closed and also 'cos I'm not specifically using phyloseq, opening a new issue for the following

Applying data transformations...
Selecting model with pulsar using stars...

 *** caught segfault ***
address (nil), cause 'unknown'

I'm trying to run SE as follows:

# Load the count file into R
prokaryote <- read.csv("prok.csv", check.names = FALSE, row.names = 1)
mat_prok <- as.matrix(prokaryote)

eukaryote <- read.csv("euk.csv", check.names = FALSE, row.names = 1)
mat_euk <- as.matrix(eukaryote)

SE <- spiec.easi(list(mat_prok, mat_euk), method = "mb", nlambda = 40, lambda.min.ratio = 1e-2, pulsar.params = list(thresh = 0.05))

I'm also attaching the input files for a cross-domain interaction network. Thanks for your help with this! prok.csv euk.csv

[susheelbhanu]

Hey @zdk123 Any insights/help into this? Thank you!

[zdk123]

still happening?

[susheelbhanu]

Yup, still has an error if I run with 'mb'. Takes forever with 'glasso' but works in the end.

[rishibhandari63]

Any update on this issue? I am also getting the same error.

[zdk123]

@rishibhandari63 can you share your data/workflow? Private email if needed.

[susheelbhanu]

I eventually went with the 'glasso' method which ran okay. The 'mb' was giving me issues but I didn't find on a test run much difference between the two methods, so went with 'glasso'.

Hope it helps!

[rishibhandari63]

@zdk123 I have shared the dataset in the email. @susheelbhanu I tried glasso but got the error asking me to use method mb instead of glasso. I am rerunning them with glasso and will update the result.

[rishibhandari63]

When I ran the network with flasso method and got the following error.

Run SpiecEasi and create association matrix for group 1

spiec_result_gr1 <- multi.spiec.easi(list(data_legume_bacteria, data_legume_fungi),

•                                  method='glasso', nlambda=40, 
  

•                                  lambda.min.ratio=1e-2, 
  

•                                  pulsar.params = list(thresh = 0.05))
  

Applying data transformations... Selecting model with pulsar using stars... Fitting final estimate with glasso... done Warning messages: 1: In spiec.easi.list(datalist, method = method, sel.criterion = sel.criterion, : input list contains data of mixed classes. 2: In pulsar(data = X, fun = match.fun(estFun), fargs = args, thresh = 0.05, : Optimal lambda may be larger than the supplied values

getStability(spiec_result_gr1) [1] 0 saveRDS(spiec_result_gr1, "legume_spieceasi_network.rds") assoMat1 <- SpiecEasi::symBeta(SpiecEasi::getOptBeta(spiec_result_gr1), mode = "ave") Error in getOptX.pulsar.refit(est, "beta") : Run spiec-easi with method="mb" Calls: <Anonymous> ... nrow -> <Anonymous> -> getOptX -> getOptX.pulsar.refit Execution halted

[rishibhandari63]

@zdk123, any update on the issue?