McGranahanLab/TcellExTRECT

negative values of TCRA.tcell.fraction

Closed this issue · 2 comments

Dagsbio commented

Hi all,

Looking into the output table of runTcellExTRECT, I see I have some negative values for the TCRA.tcell.fraction column:
image

Is this normal? Are they 0s in fact?

Thank you & Best regards,

David.

rbentham commented

Hi David,

Thank you for your comment.

Yes these are indeed 0s, there is a known issue in the package with floating point errors (Issue #17) causing very small negative fractions where the solution should be 0.

The confidence interval lower and upper values are based on the noise in the model, if the lower value is negative it should be interpreted as there being high uncertainty if there are T cells present or they are at such a low fraction they can not be reliably distinguished from the noise in the data.

Best,
Bobby

  • 👍1
Dagsbio commented

Great, thank you for the explanation Bobby.

Best,

David.