Recent technological and theoretical advances in the field of systems biology are
now guiding more systems-level analyses of immune responses.
Such analyses are particularly useful for studies in humans because of the enormous
One enabling technology is mass cytometry (CyTOF 2; Fluidigm),
which allows simultaneous profiling of all the different immune cell populations present in peripheral blood.
Novel multi-parameter measurements also requires high-dimensional data analysis to
translate such data into biological understanding.
In this project we are building a systems-level analysis on longitudinally sampled data during immune regeneration by mass cytometry.
Especially, we are building an analysis approach based on deep learning,
which is able to classify the samples and extract the significant features.
Such novel approach will highly improve the efficiency of using the high-dimensional data generated by mass cytometry,
which in turns help us to understand the underlying mechanisms that regulate immune responses in humans.
We currently have successfully built such a model to classify mass cytometry data.
However, our computing resource slows down the next step about tuning/extending the model as well as the features extraction.
Currently one training spends 4~5 days on our cpu server.
Therefore we would like to use GPU node to speed up this prosdure.