Hi Alice, I am not expert on this issue.

However, I gave a look at the manual and I found that is is possible to export to AAC (https://manual.audacityteam.org/man/aac_export_options.html). However, I suspect that you have to separately download ffmpeg in order to do that.

Hi again,

apologies for asking again, dear Fabio.

I was trying on my original data set and am getting the same LD decay and LD distance which is 0.22 and 1 , which is not correct. I am dealing with hexaploid wheat with 2223 individuals, it doesnt matter if i calculate the whole genome LD or separately for each A,B and D genome the final decay and distance remains the same.

Do you have any suggestions regarding this?

If you say, I can share my LD results from TASSEL with you to take a look, if possible.

Once again thank you very much and I really appreciate your help and kindness.

Massub.

]]>Thank you Fabio for the quick response.

I was able to complete the analyis.

1. I removed all the NaN.

2, Converted the negatives into absolute values.

3. HW.st<-c(C=0.1) instead I used HW.st<-c(C=0.01) ; It was a sample data.

Thank you very much for your help though I just have a quick question, suppose If we have large number of markers lets say 20 thousand, is there a code to thin them down for rapid analysis as otherwise it will take a lot of hard disk space.

Appreciate your help.

Massub.

]]>The error (which I guess you got from running the nls command) denotes problems in the non-linear regression. I tried hard to reproduce the error but not having the data I wasn’t able. However, I can give you some general advice:

1) NaN of LD values are useless. Remove them, and remove the corresponding distance.

2) Negative distance makes no sense. It is negative only because the arbitrary direction you used for the calculation but it is always a positive number of base pairs, so you should always use the absolute value of your distance.

3) What starting value (LD.st) did you use? By reading online I found that “illegal” starting values can generate this error, so maybe you can try to change the LD.st value and see if it helps.

I hope that by applying this suggestions your problems are solved!

Just to add here I used n=3600 as there were 600 hexaploid individuals. thank you.

]]>Thank you for this nice tutorial.

I tried to use the script for data obtained from TASSEL.

The data contained NaN for R2 and negative value for distances in some instances however, I did not change anything and got he following error

Error in numericDeriv(form[[3L]], names(ind), env) :

Missing value or an infinity produced when evaluating the mode

Can you please let me know what does it mean and how to proceed from here?

Thank you in advance.

Massub.

]]>Mmmm… I would need to see the plot to give a more detailed answer. However, the general answer is: yes, in several cases we expect the empirical and estimated thresholds to be similar. However, it is possible, if the empirical decrease is very steep or very gentle, that the two are not so close. I sometime report both the empirical value and the value estimated by the decay function.

]]>I’m using your half.decay function with a threshold of 0.2, and it gives me the location of the distance that I’m searching for.

However, the distance provided by the function is quite different from the 0.2 threshold visually observed in the plot.

They should be similar, shouldn’t they?

Thanks in advance!

]]>Hi vin.

I am sorry, I think my explanations were not so clear. When you do a LD study you usually have a number i of individuals (study subjects) for which you estimate LD between SNPs. If the species you are studying is diploid, then n is twice the number of individuals (n=2*i), because each individual contributes 2 alleles.