Normalization of data from replicates

[sylestiel]

@rcorces
Hi Ryan,

When running getMarkerFeatures(proj) on replicates for a given treatment I obtain big differences when querying for specific markers using plotMarkerHeatmap(). This is confounding the inference from different treatments.

se <- getMarkerFeatures(ArchRProj = proj, groupBy = "Sample", useMatrix = "GeneExpressionMatrix", ### bias =c("TSSEnrichment", "log10(Gex_nUMI)"))

Outside of bias adjustments what can be done to address this issue? Can you help with some suggestions.

Thank you!

[rcorces]

I'm not sure I understand your issue. Perhaps you can read this previous post and clarify.
#1409

[sylestiel]

@rcorces

Despite running
proj <- addHarmony(
ArchRProj = proj,
reducedDims ="LSI_Combined",
name = "Harmony",
groupBy = "Sample"
)

when i look to plot specific markers using the Gene Expression or Gene Score Matrix I see a lot of sample to sample variation. As seen here

[rcorces]

is your implication that these values are incorrect?

[sylestiel]

Well, I assumed that by running Harmony the replicate samples would show more of a conformity in expression whether we query for GeneScores or GeneExpression.
Is there some way to tweak this function besides using the bias =c("TSSEnrichment", "log10(Gex_nUMI)")) or bias =c("TSSEnrichment", "log10(nFrags)")) parameters?

Your thoughts/insights/suggestions would be much appreciated.

Thanks!

[rcorces]

Harmony does not alter any of the underlying data, just the positioning within the dimensionality reduction. I dont expect that adjusting any arguments will solve this.

[sylestiel]

Okay. Thanks!