CAMTHC
Posted: Tue Aug 21, 2018 4:30 pmHi Everyone,
I have come across a really amazing R based package called CAMTHC which stands for Convex Analysis of Mixtures for Tissue Heterogeneity Characterization. It uses mathematical modeling techniques to identify cellular subgroups in given samples, thus pointing towards sources of heterogeneity in a given RNASeq sample.
I am still playing around with it and trying to figure out what some of the outputs really mean - ie. when I used 3 samples, it clustered the data into 3 groups, and when I used 5 samples, it clustered the data into 5 groups etc. - which is not the output one would hope for.
Ideally, if one were to input only one heterogenous sample, the package should be able to identify with a certain amount of confidence the number of sources contained within.
For example, if one were to mix five different cell lines in one test-tube in the lab, extract the RNA, do RNASeq on it, and obtain raw expression values which would then be input into this package, it should give a graphical output with five nodes corresponding to the five different cell line sources contained within the single heterogenous sample.
I am posting here the link here, and also the relevant paper. Happy discovering! If anyone gets this to work for themselves, please post here...
Link: https://bioconductor.org/packages/devel ... AMTHC.html
Best,
Kaumudi
I have come across a really amazing R based package called CAMTHC which stands for Convex Analysis of Mixtures for Tissue Heterogeneity Characterization. It uses mathematical modeling techniques to identify cellular subgroups in given samples, thus pointing towards sources of heterogeneity in a given RNASeq sample.
I am still playing around with it and trying to figure out what some of the outputs really mean - ie. when I used 3 samples, it clustered the data into 3 groups, and when I used 5 samples, it clustered the data into 5 groups etc. - which is not the output one would hope for.
Ideally, if one were to input only one heterogenous sample, the package should be able to identify with a certain amount of confidence the number of sources contained within.
For example, if one were to mix five different cell lines in one test-tube in the lab, extract the RNA, do RNASeq on it, and obtain raw expression values which would then be input into this package, it should give a graphical output with five nodes corresponding to the five different cell line sources contained within the single heterogenous sample.
I am posting here the link here, and also the relevant paper. Happy discovering! If anyone gets this to work for themselves, please post here...
Link: https://bioconductor.org/packages/devel ... AMTHC.html
Best,
Kaumudi