Error when running quick_simulation

[EmMaKaa]

Hi,

I'd like to run SigMA on a novel gene panel, have I understood correctly that it can be done by following the example in test_determine_cutoff.R and in the Wiki page Optimization? I have tried to do that, but when running quick_simulation, I get the following error:

Error in $<-.data.frame(*tmp*, "pass", value = logical(0)) : replacement has 0 rows, data has 1

Here's how I have done it:

data_dir <- 'directory to the folder containing the vcf file'

genomes_matrix <- make_matrix(data_dir, file_type = 'vcf', ref_genome_name = 'hg38')
genomes <- conv_snv_matrix_to_df(genomes_matrix)
genome_file = 'output directory'

write.table(genomes,
genome_file,
sep = ',',
row.names = F,
col.names = T ,
quote = F)

tumor_type <- 'ovary'
remove_msi_pole <- T

data_val <- find_data_setting(genome_file,
tumor_type,
remove_msi_pole = remove_msi_pole)
data_val
##[1] "seqcap_probe"

simul_file <- quick_simulation(input_file = genome_file,
tumor_type = tumor_type,
data = data_val,
remove_msi_pole = remove_msi_pole,
catalog_name = 'cosmic_v3p2')

I haven't been able to solve this issue, am I missing something important here? Also, could you explain the difference between catalog_name options?

[dgulhan-bio]

Is there any chance run_SigMA was set to F in the quick_simulation function? It is not shown above, and by default, it should be T, but that would explain the error you get. I made some updates adding new error messages that might help resolve the issue. Would you mind rerunning and sending the error message? I also updated the test_determine_cutoff.R.

You can also try test_tune_example.R which should provide a better performance than using an existing classifier and just adjusting the cutoff. In test_tune_example.R you are generating a new classifier suitable for your data.

For obtaining the data frame that contains catalog signatures you can do:
get_catalog('cosmic_v3p2')
I also added further information on the wiki.

[dgulhan-bio]

I forgot to mention to please check out the dev branch.

[EmMaKaa]

I was able to solve this, the problem was that I was testing the method only with one test sample, which obviously didn't work (tip for other newbies here...). I will now try retuning as you recommend, thank you for the additions in the wiki, I found them really helpful.