Hi,
I try to use the Peaky with my dataset but it generates the error message followed by warning message:
"Error in if (dv > olddv && itn >= 2 && auto == TRUE) { :
missing value where TRUE/FALSE needed
In addition: Warning message:
In pnbinom(q, size = 1/sigma, mu = mu, lower.tail = lower.tail, :
NaNs produced"
I use the Peaky on a large subset, however, I set the max distance 1Mb to obtain high variance of N.
I tried to build the model using all interactions (it also ends with the error), but for purpose of visualisation I set the subsample_size=1000.
I run separately the model for each bin as multiple times Peaky output the error while computing the model for different bins using the code: "models = by(BI_fly$interactions, BI_fly$interactions$dist.bin, model_bin, subsample_size=1000)".
As the bin No 1 has the highest variance of N (please see the output below) I decided to run the model multiple times on the bin No 1 with different results. Either it fails (generating the abovementioned error) or finish with success. It is understandable that this depends on the selected subsed but what leads to the generation of the error?
Could you please help me to solve this issue?
BI_fly = readRDS(file = file_path)
#extract BIN and interaction table from BI_fly object
interactions <- BI_fly$interactions
bins <- BI_fly$bins
#extract bin_No_1
bin_1 <- BI_fly$interactions[dist.bin==1,]
bin_1$dist <- abs(bin_1$dist)
message("Build model for bin: 1")
## Build model for bin: 1
models = model_bin(
bin_1,
subsample_size=1000
)
## A truncated family of distributions from NBI has been generated
## and saved under the names:
## dNBI.0tr pNBI.0tr qNBI.0tr rNBI.0tr NBI.0tr
## The type of truncation is left
## and the truncation parameter is 0
## 18-01-2023 16:31:50
## Subsampling 1000/2851024 interactions for the null model regression...
## 18-01-2023 16:31:50
## Fitting with a maximum of 200 iterations...
## Using formula:
## N ~ log(abs(dist)) + b.trans_res + p.trans_res + sqrt(b.length) + sqrt(p.length)
## GAMLSS-RS iteration 1: Global Deviance = 5206.992
## GAMLSS-RS iteration 2: Global Deviance = 5176.838
## GAMLSS-RS iteration 3: Global Deviance = 5169.337
## GAMLSS-RS iteration 4: Global Deviance = 5167.785
## GAMLSS-RS iteration 5: Global Deviance = 5167.566
## GAMLSS-RS iteration 6: Global Deviance = 5167.539
## 18-01-2023 16:31:51
## Converged: YES
## Iterations: 6
##
## Coefficients:
## (Intercept) 8.67743046425092
## log(abs(dist)) -0.725532500725227
## b.trans_res 0.584344117331707
## p.trans_res NA
## sqrt(b.length) 0.000720678557023248
## sqrt(p.length) 0.00727315178719011
## 18-01-2023 16:31:51
## Obtaining normalized randomized quantile residuals for the full dataset...
message("Finished!")
## Finished!
Hi, I try to use the Peaky with my dataset but it generates the error message followed by warning message:
I use the Peaky on a large subset, however, I set the max distance 1Mb to obtain high variance of N.
I tried to build the model using all interactions (it also ends with the error), but for purpose of visualisation I set the subsample_size=1000.
I run separately the model for each bin as multiple times Peaky output the error while computing the model for different bins using the code: "models = by(BI_fly$interactions, BI_fly$interactions$dist.bin, model_bin, subsample_size=1000)".
As the bin No 1 has the highest variance of N (please see the output below) I decided to run the model multiple times on the bin No 1 with different results. Either it fails (generating the abovementioned error) or finish with success. It is understandable that this depends on the selected subsed but what leads to the generation of the error?
Could you please help me to solve this issue?
Check bins from BI object
Check interaction table from BI object
check the variance of the reads in each BIN
First 100 rows are displayed
BIN_2
BIN_3
BIN_4
BIN_5
Create the model for 1st bin
Session info