shajoezhu
commented
3 years ago Hi Gustovo
Thanks for the email. I am a bit busy this week. Will take a closer look at this this weekend
Best Joe
On Wed, 14 Jul 2021, 22:10 Gustavo A. Ballen, @.***> wrote:
Hello,
We have been simulating phylogenetic networks with SiPhyNetwork in R, and are trying to use hybrid-Lambda to simulate gene trees on these. However, when using some of our level-2 networks as input for hybrid-Lambda we get (1) segmentation faults, (2) a warning on lack of ultrametricity, or (3) both at the same time.
We have used the approach below for network generation:
script.R
library(SiPhyNetwork) library(ape) set.seed(10)
simulate networks under the birth-death hybridization processnumbsim = 500n1 <- 15bdNS <- SiPhyNetwork::sim.bdh.taxa.ssa(
n = n1, numbsim = numbsim, lambda = 0.9, mu = 0, nu = 0.03, hybprops = c(0.5, 0.25, 0.25), hyb.inher.fxn = SiPhyNetwork::make.beta.draw(1, 1), frac = 1, mrca = FALSE, complete = TRUE, stochsampling = FALSE, hyb.rate.fxn = NULL, trait.model = NULL )here some code testing whether the networks are level-1 or notout <- # indices to networks that are not level-1
check that all non-level-1 networks are ultrametric using apeif (sum(sapply(X = bdNS[out], FUN = ape::is.ultrametric) == length(out))) {
cat("All networks to be written to nets_not_level1.extnex are ultrametric\n")}
write all the networks above level-1 to a file (all of them are)SiPhyNetwork::write.net(net = bdNS[out], file = "nets_not_level1.extnex")
The content of net_not_level1.extnex is found attached to this issue.
Then, we use PhyloNetworks for conversion of format between extended newick and hybridlambda and write them to the file nets_not_level1.hybridlambda (also available in the attachment):
script.jl
using PhyloNetworks nets_not_level1 = readMultiTopology("nets_not_level1.extnex")
bulk format conversion of networks read in nets_not_level1open("nets_not_level1.hybridlambda", "w") do io
for i = 1:length(nets_not_level1) hyblambFormatted = hybridlambdaformat(nets_not_level1[i]) write(io, "$hyblambFormatted\n") flush(io) endendAnd finally iterate over each network to simulate a set of three loci with some arbitrary parameters:
script.sh#!/usr/bin/bash
run the r script
Rscript script.r
run the julia script
julia script.jl
iterate over the content of nets_not_level1.hybridlambda and# run hybrid-Lambda on each line in order to check issues# with ultrametricitywhile read linedo
let LNUM+=1 echo "Processing network in line $LNUM:" hybrid-Lambda -spcu $line -num 3 -seed 2 -o level2_2done < nets_not_level1.hybridlambdaThe above is run redirecting std out and err to a file (also attached):
./script.sh > outerr.txt 2>&1
The whole procedure returns 63 networks stored in nets_not_level1.hybridlambda, some of which show the problems already mentioned:
- segfault networks: 5, 6, 14, 16, 20, 27, 28, 29, 32, 36, 40, 43, 44, 50, 56, 59, 63
- ultrametric and segfault networks: 34, 52, 57, 58, 61
We would like to know if we are missing something on the way hybrid-Lambda should be run, e.g., whether level-2 networks are not guaranteed to work.
Also, we would like to understand better the warning about lack of ultrametricity, and whether it is safe to ignore it due to the fact that the simulated network is by definition ultrametric, and tested positive with ape.
Finally, we would like to understand why the segfault is happening sometimes, and whether it can be solved somehow.
Thanks in advance for your time and feedback.
Best wishes,
Gustavo @. https://github.com/gaballench), Carlos @. https://github.com/acosta187), and Claudia @.*** https://github.com/crsl4). nets_not_level1_extnew.txt https://github.com/hybridLambda/hybrid-Lambda/files/6818940/nets_not_level1_extnew.txt nets_not_level1_hybridlambda.txt https://github.com/hybridLambda/hybrid-Lambda/files/6818941/nets_not_level1_hybridlambda.txt outerr.txt https://github.com/hybridLambda/hybrid-Lambda/files/6818942/outerr.txt
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/hybridLambda/hybrid-Lambda/issues/42, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA4FP7INMIUERZVWXAG5UKLTXX4KRANCNFSM5AMHOUIA .
gaballench
Hello,
We have been simulating phylogenetic networks with
SiPhyNetworkin R, and are trying to usehybrid-Lambdato simulate gene trees on these. However, when using some of our level-2 networks as input for hybrid-Lambda we get (1) segmentation faults, (2) a warning on lack of ultrametricity, or (3) both at the same time.We have used the approach below for network generation:
The content of
net_not_level1.extnexis found attached to this issue.Then, we use
PhyloNetworksfor conversion of format between extended newick and hybridlambda and write them to the filenets_not_level1.hybridlambda(also available in the attachment):And finally iterate over each network to simulate a set of three loci with some arbitrary parameters:
The above is run redirecting std out and err to a file (also attached):
The whole procedure returns 63 networks stored in
nets_not_level1.hybridlambda, some of which show the problems already mentioned:We would like to know if we are missing something on the way hybrid-Lambda should be run, e.g., whether level-2 networks are not guaranteed to work.
Also, we would like to understand better the warning about lack of ultrametricity, and whether it is safe to ignore it due to the fact that the simulated network is by definition ultrametric, and tested positive with
ape.Finally, we would like to understand why the segfault is happening sometimes, and whether it can be solved somehow.
Thanks in advance for your time and feedback.
Best wishes,
Gustavo (@gaballench), Carlos (@acosta187), and Claudia (@crsl4). nets_not_level1_extnew.txt nets_not_level1_hybridlambda.txt outerr.txt