Open dpcook opened 3 years ago
Hi David,
Thanks a lot for your feedback and for pointing this out. First of all, thresholds are the total number of top-ranked genes in that archetype, according to ace$unified_featured_specificity, that resulted in the most significant score. We are using a novel nonparametric enrichment method that was developed after ACTIONet original submission, thus is not in the main paper, but the implementation is here
Two points to mention: 1) scale is the natural log and all computations are performed in the log-space to avoid numerical under/over-flow, 2) These are technically upper-bounds on the p-value, which is computed using a closed-form analytical solution.
for the map.cell.scores.from.archetype.enrichment() function, it just needs the annotation x archetype enrichment from any type of enrichment analysis (pathway, markers, etc.)
Please let me know if you have any additional questions
On Tue, Oct 27, 2020 at 8:01 AM David Cook notifications@github.com wrote:
Hi there--fantastic package! Really love its approach.
I'm just going through the vignette and have noticed that the gene set scoring functions may have changed since the vignette. I'm hoping to get geneset enrichment for each archetype along with cell-level scores for each.
Vignette for pathway activity scores:
Pathway.Enrichment = assess.geneset.enrichment(ace = ace, associations = nanoStringDB_human$Immunology) cell.enrichment.scores = map.cell.scores.from.archetype.enrichment(ace, Pathway.Enrichment)
assess.geneset.enrichment() doesn't seem to exist anymore though. However, I think this has moved to assess.geneset.enrichment.from.archetypes(), which produces a list containing: $logPvals (is this -log10 or -log?), which has the pathway x archetype p-values, which is great $thresholds $scores a genes x archetypes matrix
Does map.cell.scores.from.archetype.enrichment() just want $logPvals, or is it looking for a pathway score x archetype matrix from some function??
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/shmohammadi86/ACTIONet/issues/2, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABYB6H2C3IZW2AD7DOL3QBDSM3OFNANCNFSM4TA6O7EQ .
This is great--thank you! One follow up: what function are you using to generate an annotation x archetype activity score matrix? I don't think I found one, but may have missed something.
Thanks!
Hi there--fantastic package! Really love its approach.
I'm just going through the vignette and have noticed that the gene set scoring functions may have changed since the vignette. I'm hoping to get geneset enrichment for each archetype along with cell-level scores for each.
Vignette for pathway activity scores:
assess.geneset.enrichment()
doesn't seem to exist anymore though. However, I think this has moved toassess.geneset.enrichment.from.archetypes()
, which produces a list containing:$logPvals
(is this -log10 or -log?), which has the pathway x archetype p-values, which is great$thresholds
$scores
a genes x archetypes matrixDoes
map.cell.scores.from.archetype.enrichment()
just want$logPvals
, or is it looking for a pathway score x archetype matrix from some function??