Why doesn't the regression lines intersect the y-axis (dose=0) at the measured 0Gy-control measurement?
By definition, the Surviving fraction should be 100% if no treatment was performed
with the cells.
So after normalisation, the graph should intersect Y-axis at value 1 where dose=0.
But what if the colony counts for untreated controls where inexact?
Instead of using the experimental - potentially incorrect - control colony numbers,
we can estimate the colony number at dose=0 from the regression.
Each single dose implicitely contains the untreated colony number and thus the regression value shold supply
a more correct estimate of the dose=0 value, but it might slightly deviate from the measured value (which might be inaccurate).
My graph for photons bends up - Why?
The line in the graph represents the best fit of the linear-quadratic model to your data.
If the "line" has an unexpected shape, this is due to a problem within your entered data.
Most probably:
the number of counted colonies might be inaccurate
numbers of seeded cells for the different doses was mixed up