Flow cytometry and Cytof data analysis

Question

[rashmirao962]

What is the linear region and co-factor ? How are they related?

https://rdrr.io/bioc/diffcyt/man/transformData.html

[rashmirao962]

@Kevin Blighe , thank you for the answer. My understanding is that, (1)The linear region is the area under the curve,not the area of the bar with the highest frequency. (2)with increase in co-factor the width of the linear region decreases. (3)The purpose of the transformation is to scale down the data and achieve a more Gaussian distribution of the data set.However, in the PBMC dataset(http://imlspenticton.uzh.ch/robinson_lab/cytofWorkflow/).Upon plotting the histogram for different co factor values(2,5,10)for the arcsinh transformed expression matrix ,I noticed that at at co-factor value of 2 the distribution is more Gaussian than at co-factor value of 5. (4)Wouldn't 2 be more suitable in this case?

Co-factor 10

Co-factor 5

Co-factor 2

Thank you,

Rashmi

Answer 1

[Kevin Blighe]

Hi, I think that the function documentation explains it quite well, and in a simple way.

Key points:

1. The inverse hyperbolic sine ('arcsinh') transformation is widely used in cytometry
2. arcsinh is similar to log transformation at high values, but is 'linear' with values near zero. So, unlike log, arcsinh can work with zero values, and even small negative values
3. This arcsinh transformation comes with one key adjustable parameter: 'cofactor'
4. Cofactor controls the 'width' of this linear region
5. Recommended values for cofactor are 5 for mass cytometry and CyTOF, or 150 for flow cytometry

Kevin