Entering edit mode
I have a simple question on what is inflection and knee points on the DropletUtils barcode rank plot means? What is the difference?
Also, how is the lower bound threshold set for this?
I am new to single-cell rna-seq and I am confused about this concepts.
The knee is the point where the curve (rank vs total) makes a sharp drop, used as a cutoff towards which barcodes we believe are 'real' versus spurious. You can google what an inflexion point is mathematically. In essense, the inflexion point is used to narrow down the range of x-axis values that are used to fine the knee. Source: A GitHub issue at DropletUtils that I cannot find right now but remember reading it. Can be found via google for sure.
thank you! does this mean that the cells above the knee point are "real" and below the knee point are "empty droplets"?
Empty droplets, poorly-captured cells, contamination. Any barcode found at low and non trustworthy frequency. Many articles online on the knee method in sc data.