devillemereuil/bayescenv

BayeScEnv is a Fst-based, genome-scan method that uses environmental variables to detect local adaptation.

bayescenv's Issues

Segmentation fault (core dumped)

Hi
When I used the bayescev to analysis my dataset, and get the error: Segmentation fault (core dumped) ./bayescenv-1.1/bin/linux64/bayescenv data.txt -env bio.txt -o test -threads 16 -o test -nbp 10 -pilot 2000 -thin 10 -n 5000 -burn 10000

This is my part of data.txt
[loci]=449306

[populations]=18

[pop]=LA1958
1 20 2 9 11
2 20 2 11 9
3 20 2 20 0
4 20 2 9 11
5 20 2 20 0
6 20 2 17 3
7 20 2 14 6
8 20 2 20 0
9 20 2 20 0
Is it because of too much data?
I try to run the test data data_codominantSNP.txt with my env data, and it is OK.

not enough memory

Hi I am facing an error with bayescenv...

My Verif file only has these lines:

"Summary of parameters and input files. Please check that all is correct while calculation is starting... There are 15947776 loci. There are 2 populations."

There is no error message. And the log file only says:

"This is BayeScEnv version 1.1. Using 20 threads (20 cpu detected on this machine) Not enough memory for pop[0].locus"

and the program exits right after this.

.sel program is empty
and .Fst is not generated.

I imagine there is something to do with the number of SNPs in my input, which is big because I am dealing with bird whole genomes.

Thank you so much for any help!!!
Best wishes

q-values of 0

I am running bayescenv on a large SNP dataset, and manhattan plotting -log10(q-value) to visualize significance across the genome. I have now realized that my most significant SNPs are not being plotted because their q-value is 0, so the -log10 becomes infinity. I have manually converted the 0 qvalues to 1e-5, so that they will plot, but I just wanted to verify that this is not a bug, and ask what you would recommend in this case. Thanks for your time!

How to get the Environmental data

Hi I still do not kown how to get the Environmental data. Can you help me?

Running calculation

Hi,

I am using BayeScEnv-0.3 with Mac (ver.10.8.5) and trying to analyse for detecting outliers which associated with environmental variables.
I am now using test files that are in a directory called "test" which was included in bayescenv-0.3 package. but when I started to run the analysis under terminal, there seems to be an error and says "Abort trap: 6". and then I also used GUI, calculation freezes in the middle of analysis.
What might be a problem here? is that my computer? or the command I used was incorrect?
I use a command in below
./bayescenv data.txt -env env.txt -out_pilot -all_trace -unif_pr 10 -pr_jump 0.1 -pr_pref 0.5 -o test -burn 3000 -thin 10 -nbp 10 -pilot 2000

Thanks in advance for any suggestions.

Ayako

huge data set

Hi @devillemereuil ,

I'm currently running BayeScEnv on my WGS data, considering such a large data set (1 million sites), is it feasible to divide them into multiple subsets？
Actually I have tried with the command "-nbp 10 -pilot 2000 -thin 10 -n 5000 -burn 10000 ", but when calibrated the MCMC, I found that the results didn't reach convergence. So I increased the parameters ("-nbp 20 -pilot 5000 -thin 10 -n 5000 -burn 20000"), but it takes a very long time to run. Thus I'm wondering if I could divide my data into multiple subsets.
Thanks in advance.

Best,
yuxi Hu

AFLP data yield low q-values for g

When using AFLP dataset (e.g. data_dominantAFLP.txt), the q-values yielded for g are very low when they should be close to 1.

Input file paths with spaces breaks the GUI

When using file path locations with spaces, the GUI states "Could not open file". This is due to the path not being encapsulated between quotes when the bayscenv binary is called.

PrJump and PrPref should be switched in GUI

The default probability of non neutral model (pi) should be 0.1 not 0.5, whereas the preference for the locus-specific model should be 0.5, not 0.1.

some question about the environmental file

Hi @devillemereuil ,

I am a bit confused about the environmental data. As the Wiki describes, I have computed the environmental variables as the distances, and standardized them. Then I got the environment file like this:

-1.628513663 0.282856354 0.410566501 0.310504473 -0.297343129 -1.272742484 -1.430681843 -0.909762816 1.685776976

I noticed that there were some negative values and some positive values. However, Bayescenv only consider absolute value. And when I run Bayescenv with this environment file, the output file *.Verif.txt looks like this:
Raw environmental variable:
-1.62851 0.282856 0.410567 0.310504 -0.297343 -1.27274 -1.43068 -0.909763 1.68578
Distance-transformed (i.e. absolute value) environmental variable:
1.62851 0.282856 0.410567 0.310504 0.297343 1.27274 1.43068 0.909763 1.68578

It just transferred all values to positive numbers. I am not sure whether I could just use this environment file which contains negative values or I could added the absolute value of the largest negative number to all the numbers so that the variables would all be positive.

Any help would be appreciated!

Methods question

Hi there,

Can you provide some more details about how missing data are handled in this software? I'm guessing that since the method utlimately uses the allele counts, missing data are essentially ignored.

Thank you!

Amanda

Segmentation fault (core dumped)

Hi I tried to run BayescEnv with an SNP matrix using the '-snp' tag but no matter what I will end up with the error message "Segmentation fault (core dumped) ". Even if I reduce the number of SNPs to 100 I get this error. I hope you can help me to solve my problem.

Thanks