deletions | additions       diff --git a/The model reproduce observe phase shift.tex b/The model reproduce observe phase shift.tex  index 516d7df..4431b29 100644  --- a/The model reproduce observe phase shift.tex  +++ b/The model reproduce observe phase shift.tex 
...
\subsection{Shifting phases}  A typical feature of synchronization is that the phase difference between the two oscillators depend on the difference between their intrinsic periods (TODO ref). To test whether our dataset exhibit this important characteristic we first estimated the position of the stochastic attractors from the data in three conditions with different cell cycle durations. When grown at 40\degree the cell cycle duration is significantly shorter (TODO number) than at 37\degree, the resulting attractor is shifted to the right left  in the phase plane (Figure \ref{fig:phase_shift}A, red verus black). On the contrary when grown in the presence of a mild (i.e. non-blocking) concentration of a CDK2 inhibitor, the cell cycle duration is lengthened and the attractor shifts to the right (Figure \ref{fig:phase_shift}A, blue). It can also be observed that the shift is maximal just before passing by the interaction region centered around $(0.75,1)$ after a period of free diffusion. We next quantified the mean phase shift in circadian phase at different cell cycle phase (Figure \ref{fig:phase_shift}A, gray lines) in different conditions that alter the cell cycle or the circadian intrinsic period. The resulting linear phase shifts (Figure \ref{fig:phase_shift}B, black) confirm that the cell cycle and the circadian clock are in resonance.  The stochastic model (Figure \ref{fig:phase_shift}B, red) a is also able to reproduce these shifts, although it overestimate the slope close to divisions.