What is best way to do CNV Analysis on Xenium data (Spatial transcriptomics)?

Question

[geno89]

I know about InferCNV but i dnt find i think it works on whole genome data while we have xenium sections.

[Michael]

Hi, could you please provide a little more detail on your setup? I don't even believe that copy number variation in the genome can be reliably infered from transcriptomics data even with this package available which is seemingly not published or benchmarked.

[yura.grabovska]

The methodology has been published even if the package itself as a unit has not. https://doi.org/10.1126/science.1254257

I have come across benchmarking papers in the past and in each case, the literature stresses the 'infer' aspect of the methods. These approaches are extremely wobbly in the result and require a fair amount of interpretation. Either way, I have in the past had OK results with recapitulating bulk CNV profiles from pseudo-bulked single-cell inferred CNV profiles.

Answer 1

[yura.grabovska]

What panel have you used on your Xenium section? The first thing to test would be how much of the genome you cover per chromosome. In theory, if you have enough coverage across a chromosome you could bin your genes into windows. You might not have very good resolution, but realistically you could call chromosomal, or arm-level events?