Rules of CNV/SV signature selection

[RuimeiL]

Hello, when using Sigminer for CNV (Copy Number Variation) and SV (Structural Variation) signature analysis, the selection allows me to choose n signatures. How are the confidence scores of these signatures ranked? Is there a scoring system? Typically, how many signatures are recommended for selection?

[ShixiangWang]

@RuimeiL Hi, thanks for your questions. Could you put more details?

[RuimeiL]

for example cn <- read_copynumber(segTabs_own, seg_cols = c("chromosome", "start", "end", "segVal"), genome_build = "hg38", complement = FALSE, verbose = TRUE)

cn_tally_W <- sig_tally(cn, method = "W") sig_w <- sig_extract(cn_tally_W$nmf_matrix, n_sig = ?)

n_sig = ?, I can set any number here. n_sig =1,2,3,4,5......

How are the confidence scores of these signatures ranked? how many signatures are recommended for selection? Thank you for your time.

[ShixiangWang]

Hi @RuimeiL , please roughly read the documentation https://shixiangwang.github.io/sigminer-book/index.html (PART 1 and 2 specifically) and https://cran.r-project.org/web/packages/sigminer/vignettes/cnsignature.html

If you don't have the concept of extracting signatures (signature types and basic operations, difference between signature extraction and fitting) and setting the parameters, you cannot do the correct analysis.

For your specific questions, you can find the answer at https://shixiangwang.github.io/sigminer-book/basic-workflow.html#de-novo-signature-discovery

[RuimeiL]

Thank you so much!