Wed Jun 01, 2016 9:49 pm
2016-Li et al-biorxiv
"DeepCyTOF: Automated Cell Classification of Mass Cytometry Data by Deep Learning and Domain Adaptation"
http://dx.doi.org/10.1101/054411
For the computational folks on this list. It was only a matter of time until someone did this since deep learning is hot right now.
A few initial thoughts -- not many cell subsets overall (it would be fantastic to learn more cell types), but this tradeoff is understandable given the lack of publicly available data for training. In addition, I find it interesting that multivariate (softmax) regression actually does quite well independent of the DeepCyTOF autoencoder-based framework.
http://dx.doi.org/10.1101/054411
For the computational folks on this list. It was only a matter of time until someone did this since deep learning is hot right now.
A few initial thoughts -- not many cell subsets overall (it would be fantastic to learn more cell types), but this tradeoff is understandable given the lack of publicly available data for training. In addition, I find it interesting that multivariate (softmax) regression actually does quite well independent of the DeepCyTOF autoencoder-based framework.