Illumina / Cyrius

A tool to genotype CYP2D6 with WGS data
Other
48 stars 5 forks source link

WGS only? #26

Closed JARM89 closed 2 years ago

JARM89 commented 2 years ago

Have anyone tested Cyrius with data from a chip or single sequenced gen? Thank you in advance.

xiao-chen-xc commented 2 years ago

Hi @JARM89 Cyrius is designed for WGS data only. The copy number calling algorithm uses assumptions of WGS and will need modifications to work for other types of data.

JARM89 commented 2 years ago

Thank you for the information.

BTHermann commented 2 years ago

Hi @xiao-chen-xc , can those modifications be done by specifying other reference regions in the genome? Assuming that refernce regions for CN determination are specified in CYP2D6_region_38.bed ( _19 or _37, respectivley), is it possible to customize this file accoding to a utilized (targeted) panel?

Man thanks in advance for your help.

xiao-chen-xc commented 2 years ago

Hi @BTHermann, Cyrius assumes that the coverage is uniform across the regions specified in the bed file so it only does simple normalization and GC correction. For targeted data, there is more variation across the panel, so one needs to add the logic to normalize depth based on depth patterns specific to the panel.

BTHermann commented 2 years ago

Thank you very much @xiao-chen-xc for your quick reply! I just have one follow question: when defining custom reference regions, in addition to uniformity of coverage, are there more requirements? For example, is a minimum number of reference regions required? Or does a reference region has to span a certain minimal range?

xiao-chen-xc commented 2 years ago

For WGS, they are there just to provide some median normalization and GC normalization, and they are simply some regions of the genome that have stable coverage across samples. If you want to add your own normalization method for targeted data, it really depends on your own need, i.e. you can decide the number and size of regions that would allow best normalization results.

BTHermann commented 2 years ago

Allright, thank you for your feedback! It is much appreciated.