Questions about sig_fit()

[xiw588]

Hi Shixiang,

I have a couple of questions regarding sig_fit () and hope you can help me better understand!

1) I try to run sig_fit(catalogue_matrix =mat, sig=mt_sig,type = c("absolute", "relative"),show_index = TRUE,auto_reduce=TRUE,return_class = "data.table",rel_threshold = 0). But it only gave me the results of absolute exposure rather than both absolute and relative exposures.

2) Can you please explain what the absolute exposure is measuring here? Are they referring to the number of mutations that can be attributable to different mutational signatures? If that is the case, how does the non-integer values come? For example, 6.028082.

Thanks!

[ShixiangWang]

@xiw588 Hi,

For your first question, the function only handles one type of the type, when two are input, only the first one is considered.

For your second question, you are right about the explanation. The absolute value is an estimated mutation number assigned to a signature. As it is estimated, a numeric instead of an integer is expected. You can round it if you care about the integer value.

[xiw588]

Hi Shixiang,

Thank you so much for your response! I have a follow up question - I understand that the tool is designed for WES or WGS data whereas the signature estimation procedure may not perform well for the panel data. I wonder if same issues occuring for APOBEC enrichment only focusing on t_tally$APOBEC_scores? Thanks!

[ShixiangWang]

If you are using panel data, you should be careful about the data as the unsufficient data may generate unexpected results. A tool called SigMA may help you.

[xiw588]

Thanks a lot!