search

morrislab / pairtree

Pairtree is a method for reconstructing cancer evolutionary history in individual patients, and analyzing intratumor genetic heterogeneity. Pairtree focuses on scaling to many more cancer samples and cancer cell subpopulations than other algorithms, and on producing concise and informative interactive characterizations of posterior uncertainty.
MIT License
37 stars 11 forks source link

is pairtree not also using longitudinal data? #42

Closed sahuno closed 1 year ago

sahuno commented 1 year ago

Dear developers, pardon my naiveté.

reading the star protocol it says: "......while others focus on building clone trees using longitudinal data, such as CALDER ([Myers et al., 2019]"......... "Pairtree differs from existing methods by accommodating these large amounts of data, such that it exploits information from many samples taken from the same cancer to build accurate clone trees even as the number of subclonal populations increases." .......

Que - is the assumption that you take many samples from same cancer same day? because if you take many samples across different time points it becomes longitudinal data/sampling. Or maybe i'm missing something?

as always, thank you for your help -S

ethanumn commented 1 year ago

Hi there --

Pairtree can perform reconstructions using longitudinal data, however, Pairtree does not have an explicit parameter that allows you to tell the algorithm the data is longitudinal. Pairtree instead uses the data directly to infer the most plausible evolutionary trees. There are other algorithms (e.g., CALDER) that have a parameter that you can set which will tell the algorithm that the samples are longitudinal. This setting restricts the algorithm to only search for trees that adhere to the longitudinal constraints.

Hopefully that helps clear things up! -Ethan

sahuno commented 1 year ago

thanks @ethanumn for this exposition! -S