Autotune the R parameter

[crow2009]

Hello,

Thanks for the nice tool.

I am wondering if the parameter "autotune = TRUE" also works for the backcross population (nstates = 2).

Thank you

[crow2009]

I got errors as below:

Fitting the parameters and Viterbi decoding.
post processing value is: TRUE
R value autotune is: TRUE
Fitting an RTIGER object with the R provided by the user.
Optimizing the R parameter.
Error in params[["paraBetaAlpha"]][state, 1] : subscript out of bounds
Calls: RTIGER ... optimize_R -> extract_emissions -> lapply -> lapply -> FUN

[crow2009]

Sorry for the message bombing.

When I tried without the parameter "autotune = TRUE", only use "rigidity = 200". I got errors as below.

Fitting the parameters and Viterbi decoding.
Error in cat("post processing value is:", post_post.processing, "\n") :
object 'post_post.processing' not found

Just to make it more clear, I used the GitHub version, not the released version (no "autotune" parameter).

Many thanks for your help in advance.

[rfael0cm]

no worries at all, thank you very much for testing it and giving feed back! I really appreciate it.
I have fixed the last issue (post_post.processing). This should work now perfectly.

About the autotune function for different states than 3, it should work as well, but let me check it on detail and get back to you asap.
thanks a lot!

[crow2009]

Many thanks for your swift reply.
Yes, The issue (post_post.processing) was fixed. Now, it works smoothly. Thank you.

Looking forward to your update about the autotune function for the BC1 population!

I tried the autotune function for the wild-type F2 sample (4056_U) which was used in your paper. But the result looks a bit weird. You can see below, 8 was chosen as the best R-value, and then resulted in a huge amount of COs. Could you please help to have a look?

Fitting the parameters and Viterbi decoding.
post processing value is: TRUE
R value autotune is: TRUE
Fitting an RTIGER object with the R provided by the user.
Optimizing the R parameter.
Best R value: 8
Number of iterations run: 6

Thanks a lot!

[rfael0cm]

are you using only one sample? why did it stop after 6 iterations? Is that the limit you set or has it stop by itself?

[crow2009]

Hi,

Yes. Only that sample.
I didn't set any limitations. Only used the defaults.
expDesign = , outputdir = , seqlengths = , rigidity = 10, autotune = TRUE, save.results = TRUE

Here are all the prints:

Fitting the parameters and Viterbi decoding.
post processing value is: TRUE
R value autotune is: TRUE
Fitting an RTIGER object with the R provided by the user.
Optimizing the R parameter.
Best R value: 8
Number of iterations run: 6

Plotting samples Genotypes.
PLotting CO number per chromosome.
Creating bed and IGV output formats.
Plotting goodness of fit.
Warning messages:
1: replacing previous import ‘ellipsis::check_dots_unnamed’ by ‘rlang::check_dots_unnamed’ when loading ‘hms’
2: replacing previous import ‘ellipsis::check_dots_used’ by ‘rlang::check_dots_used’ when loading ‘hms’
3: replacing previous import ‘ellipsis::check_dots_empty’ by ‘rlang::check_dots_empty’ when loading ‘hms’
Julia exit.

Thank you

[rfael0cm]

Sorry for the delay in answering! I got caught up on another project and couldn't focus on this. I have been looking into that, I did realize that there was an error, but this led to finding an even bigger error in the optimize_R algorithm. I am working on fixing it and I will get back to you asap it is done. This will also delay optimizing the R for states != 3 for the moment.
Really sorry for this!
Nevertheless, the RTIGER fitting function should work normally and get standard results. :-)

[crow2009]

Thanks. That would be cool. Looking forward to it.