search

hammerlab / SmartCount

Repository for collaboration on Celldom computer vision solutions
Apache License 2.0
2 stars 0 forks source link

Mean cell counts across arrays #61

Closed eric-czech closed 5 years ago

eric-czech commented 5 years ago

Previously, the array_analysis notebook had cell count sums across arrays but this doesn't make sense as it does not account for:

  1. Initial condition filters that should exclude certain apartments from all analysis
  2. Slightly different counts in the number of apartments detected for each array

Instead, this was refactored to show average cell counts across time for each array (with fixed and dynamic scales):

screen shot 2018-11-28 at 8 11 52 am

Additionally, this fixes the poor (and incorrect) descriptions of some parameters in the resistant_cell_analysis

yellenlab commented 5 years ago

@eric-czech , this is really great. However, I would like to plot something slightly different. In essence, it should be possible to plot a smooth analytical curve for the size of the population on chip, S(t), as a function of time, which has the benefit of predicting the population size at time points beyond what we normally would do in an experiment. For example, in the images above, we see that the population recovers more slowly at higher drug concentration. Using the expression:

S(t) = sum(2^(lambda_i*t))/N

you can extrapolate the curve to future time points to determine the TTR. We would expect that there should be a relationship between drug concentration and TTR that could yield a required drug dosage to prevent the TTR from recurring within X many years. If we are able to demonstrate this relationship, then it could have clinical significance in terms of a personalized metric for the activity of a patient's tumor against a specific drug.

@jmotschman

eric-czech commented 5 years ago

Like in https://github.com/hammerlab/celldom/issues/59#issuecomment-442551664?