Open chiaramazzoni opened 4 years ago
Can you maybe share the input files you used for Desman and VariantFilter?
Sure, here are the files from the Variants step. freqs_sel_var.txt freqs_tran_df.txt
Thanks
Desman is limited in the number of strains that can be reliably resolved to less than ten roughly. However, it looks to me like you have six samples and lots of variants so it should work I will try running it with g = 2,...,8 and see what we get.
I was able to run this without issues. However the results are perhaps not that interesting. It does look like six strains. One dominating each sample. I have attached the Deviance plot. Dev.pdf
This is the output of resolvenhap: 7,6,3,0.003041666666666667,Test_7_3/Filtered_Tau_star.csv
So six optimum strains. If you want all the results I can upload them?
I was able to run this without issues. However the results are perhaps not that interesting. It does look like six strains. One dominating each sample. I have attached the Deviance plot.
Thank you very much for helping! So just to make it clear, the presence of secondary strains in each sample is considered and then rejected in the process?
> Desman is limited in the number of strains that can be reliably resolved to less than ten roughly. However, it looks to me like you have six samples and lots of variants so it should work I will try running it with g = 2,...,8 and see what we get.
Apologies for hijacking this threads. Based on the above quoted statement, can I ask A) if it is okay to use DESMAN when one knows that there are indeed more than 10+ strains exists across 400+ metagenomes? or B) DESMAN would simply not reliably predict anything (roughly) above ten strains?
Many thanks, Shriram
It always stopped in ‘Start Gibbs sampler burn-in phase’ in the log file, what's the problem?
Hello,
I'm following the complete example on E.coli strains but applied to a species of Lactobacillus. Everything is fine and outputs seem reasonable till running
desman
for inferring strains..I ran it with-r 1000 -i 100
( but also-i 500
) and it finishes withNTF Iter 1800, div = 0.009128
andStart Gibbs sampler burn-in phase
with empty outputs. I know that at least from my co-assembly there should be at least one strain per sample, so I set g at least equal to n° of samples. I know there are snps on core genes that I identified because I separately called bcftools on the contigs.Thanks in advance for the support !