Comparing Sequenza output to FACETS output

[Sviatoslav Kendall]

I'm relatively new to analyzing copy-number variations and have mainly used data produced using FACETS in the past. I'm interested in comparing a dataset with copy number data produced using FACETS to a dataset produced using Sequenza. I do not have access to the BAM files so I cannot simply re-run the analysis using the same software. 

Wondering if anyone has experience working with both of these types of CNV data who can comment on how much sense it makes to do this kind of analysis and what sorts of pitfalls to watch out for. 

[Francesco Favero]

Hi,

it looks like a fun idea to convert the file format for FACETS in a "seqz", how the input files was generated? Did it includes tumor and normal pairs?

Francesco

Il 21 mar 2017 15:58, "Sviatoslav Kendall" <sviatosla...@gmail.com> ha scritto:

I'm relatively new to analyzing copy-number variations and have mainly used data produced using FACETS in the past. I'm interested in comparing a dataset with copy number data produced using FACETS to a dataset produced using Sequenza. I do not have access to the BAM files so I cannot simply re-run the analysis using the same software. 

Wondering if anyone has experience working with both of these types of CNV data who can comment on how much sense it makes to do this kind of analysis and what sorts of pitfalls to watch out for. 

--
You received this message because you are subscribed to the Google Groups "Sequenza User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sequenza-user-group+unsub...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Sviatoslav Kendall]

Both the FACETS and Sequenza datasets are tumor-normal pairs that were analyzed using whole-exome BAM files. 

[Francesco Favero]

Ah, sorry, I was going thinking you didn’t have the data to run sequenza, only the FACETS input.

I've never tried FACETS, as it is quite recent, and they didn’t include a comparison with sequenza in the manuscript, It would be nice indeed to hear someone that have some experience with both, or read a comparison between the two tools (maybe among others).

From the FACETS manuscript I can see that there are many overlaps between the two packages, they are basically designed to do the same thing, FACETS is a very new addition to the "tumor-SCNA caller”, so it may cover more specific needs, such as the data from sequencing panel (I’ve used sequenza with sequencing panel data, so that’s not impossible), and provide sublclonal heterogeneity information, which sequenza at its current release doesn’t account for.

Regarding the pre-processing of sequencing data, sequenza genotypes the samples by looking at the base content of the normal sample, while FACETS can select SNP from dbSNP and the 1k genome projects.

Sequenza uses an external package to perform an allele specific segmentation (copynumber), while FACETS implement a new method to do so. So the segments resulting from the two tools may be completely different (the segmentation is the most challenging task, in my opinion). So the processed input data and the segmented data can be very different between the two tools.

They both uses depth ratio and the “B allele” frequency, although FACETS uses the log-transformed data for both.

The ploidy/cellularity model seems similar, The main differences are in the implementations (obviously) and also a quite different approach on the inference of the parameters.

The cellularity/ploidy solution can be selected manually in sequenza. I’m not sure if FACETS present such choice. So a lot depends if the two datasets have been “visual inspected” or run both unsepervised.

In practice the two tools should give equivalent results, I’ve never used FACETS, but they it looks like it does provide segmented data with integer and allele-copy number results as sequenza does, so a comparison between two dataset should be doable. 

Francesco

To unsubscribe from this group and stop receiving emails from it, send an email to sequenza-user-g...@googlegroups.com.