search

HCBravoLab / metagenomeSeq

Statistical analysis for sparse high-throughput sequencing
64 stars 20 forks source link

fitZig Error #55

Closed ninaxhua closed 6 years ago

ninaxhua commented 6 years ago

Hello, I'm trying to use fitZig since it can implement Limma's duplicateCorrelation function. I received this error when I ran: fit_MRE = metagenomeSeq::fitZig(obj = n_paired_merged_MRE, mod = mod, useCSSoffset = FALSE, control = zigControl(), useMixedModel = TRUE, block = pData(n_paired_merged_MRE)$id)

Error: it= 0, nll=232.25, log10(eps+1)=Inf, stillActive=1141 it= 1, nll=232.22, log10(eps+1)=Inf, stillActive=622 it= 2, nll=235.50, log10(eps+1)=Inf, stillActive=209 it= 3, nll=237.86, log10(eps+1)=Inf, stillActive=24 it= 4, nll=237.98, log10(eps+1)=Inf, stillActive=12 it= 5, nll=238.06, log10(eps+1)=Inf, stillActive=6 it= 6, nll=238.10, log10(eps+1)=Inf, stillActive=5 it= 7, nll=238.18, log10(eps+1)=Inf, stillActive=4 it= 8, nll=238.21, log10(eps+1)=Inf, stillActive=4 it= 9, nll=238.23, log10(eps+1)=Inf, stillActive=4 Error in if (abs(correlation) >= 1) stop("correlation is 1 or -1, so the model is degenerate") : missing value where TRUE/FALSE needed

jnpaulson commented 6 years ago

Thanks - Can you provide an MRE?

Best,

On 2017-11-13 10:56, ninaxhua wrote:

Hello, I'm trying to use fitZig since it can implement Limma's duplicateCorrelation function. I received this error when I ran: fit_MRE = metagenomeSeq::fitZig(obj = n_paired_merged_MRE, mod = mod, useCSSoffset = FALSE, control = zigControl(), useMixedModel = TRUE, block = pData(n_paired_merged_MRE)$id)

Error: it= 0, nll=232.25, log10(eps+1)=Inf, stillActive=1141 it= 1, nll=232.22, log10(eps+1)=Inf, stillActive=622 it= 2, nll=235.50, log10(eps+1)=Inf, stillActive=209 it= 3, nll=237.86, log10(eps+1)=Inf, stillActive=24 it= 4, nll=237.98, log10(eps+1)=Inf, stillActive=12 it= 5, nll=238.06, log10(eps+1)=Inf, stillActive=6 it= 6, nll=238.10, log10(eps+1)=Inf, stillActive=5 it= 7, nll=238.18, log10(eps+1)=Inf, stillActive=4 it= 8, nll=238.21, log10(eps+1)=Inf, stillActive=4 it= 9, nll=238.23, log10(eps+1)=Inf, stillActive=4 Error in if (abs(correlation) >= 1) stop("correlation is 1 or -1, so the model is degenerate") : missing value where TRUE/FALSE needed

-- You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub [1], or mute the thread [2].

Links:

[1] https://github.com/HCBravoLab/metagenomeSeq/issues/55 [2] https://github.com/notifications/unsubscribe-auth/AC5hg2FZW4W3FHocPLUnIPl8Bm-B7Vi8ks5s2GatgaJpZM4Qb_D4

ninaxhua commented 6 years ago

Link to drive folder. The object is called n_paired_merged_MRE.

jnpaulson commented 6 years ago

Hi Nina,

Apologies this functionality is still under development. The output of the fitZig model does include a fitted MArrayLM object. From that object you can go ahead and use the duplicateCorrelation function.

ninaxhua commented 6 years ago

Hello,

So this is what I have so far fit_MRE = metagenomeSeq::fitZig(obj = n_paired_merged_MRE, mod = mod, useCSSoffset = FALSE, control = zigControl()) dupCor_MRE = duplicateCorrelation(fit_MRE$fit, mod, block = pData(n_paired_merged_MRE)$id)

I got a data class error when using fit_MRE$fit as an object so I'm assuming I have to use one of these values as the object for duplicateCorrelation, but I'm not sure if I would use the coefficients matrix: coefficients stdev.unscaled sigma df.residual cov.coefficients pivot rank Amean method design df