search

navinlabcode / copykat

Other
193 stars 53 forks source link

The results of diploid and aneuploid are reversed in test_copykat_heatmap.jpeg, why? #99

Open lj20220425 opened 10 months ago

lj20220425 commented 10 months ago

Sorry to bother you, when I run the following code: copykat.test <- copykat(rawmat=exp.rawdata, id.type="S", ngene.chr=5, win.size=25, KS.cut=0.1, sam.name="test", distance="euclidean", norm.cell.names="", output.seg="FLASE", plot.genes="TRUE", genome="hg20", n.cores=4) Sometimes, the results of diploid and aneuploid were logical in the test_copykat_heatmap.jpeg; while sometimes the results of diploid and aneuploid were reverse in the test_copykat_heatmap.jpeg. Which should be aneuploid , but titled with diploid. Can you tell me why this happend and how to solve it. Best wish

tlesluyes commented 8 months ago

Hi dev team,

Similar experience here, I feel like the diploid/aneuploid categories are swapped for some reason. I'm attaching a JPEG file. Please let us know how to fix this.

Cheers,

Tom. T34592_copykat_heatmap

ZhangMH2000 commented 8 months ago

I encountered the same issue. I had to reverse the results manually.

ZhangMH2000 commented 8 months ago

I found another issue: I repeated copykat for my dataset due to some test reason. The new output shows a correct result, and the heatmap was correct. However, there are some little differences between the two times, and the predicted result changed slightly in one sample(10% cells). I’m not sure whether it’s normal.

gloriafight commented 3 months ago

I had a similar problem. Some of normal cells are marked as malignant cells, and Some of malignant cells are marked as normal cells. However, through the clustering of Seurat, it can be found that malignant cells are clustered together. fufa radio_copykat_heatmap image The lower left corner is a malignant cells.

kaizen89 commented 2 months ago

I also observe results being reversed on a sample among cells from tumor vs adjacent tissues (adjacent cells being classified as aneuploid). Unexpectdly, when I run copykat on merged samples (I assume batch effect) then most of the cells of adjacent tissue are classified as diploid.