search

nolanlab / citrus

Citrus Development Code
GNU General Public License v3.0
31 stars 20 forks source link

FlowCAP-I setup #108

Closed hl475 closed 7 years ago

hl475 commented 7 years ago

Hi I have a question regarding the set-up of citrus for FlowCAP-I mentioned in the paper 'Automated identification of stratifying signatures in cellular subpopulations'.

Do you do any kind of transformation or scaling when you read the data? For example, the GvHD dataset, I did the following according to Table S2 of the paper 

clusteringColumns = c("FL1.H","FL2.H","FL3.H","FL4.H")
citrus.combinedFCSSet = citrus.readFCSSet(dataDirectory, fileList, fileSampleSize = 10000)

citrus.clustering = citrus.cluster(citrus.combinedFCSSet, clusteringColumns)
largeEnoughClusters = citrus.selectClusters(citrus.clustering,
                                            minimumClusterSizePercent = .05)

citrus.combinedFCSSet contains 12 * 10000 = 120000 events. I didn't get length(largeEnoughClusters) = 46 as said in column 3 Table S2 for GvHD. I have also tried with specifying transformColumns, transformCofactor, scaleColumns.

What do I miss here? How should I set up the experiment to get length(largeEnoughClusters) = 46 as said in column 3 Table S2 for GvHD? Thanks!

rbruggner commented 7 years ago

I believe column 3 of summary table S2 the total number of manually identified populations in the FlowCap1 GvHD dataset. That is, there are 12 files, with say 3 to 4 manually identified populations in each of the files = about 46 populations. That table does not refer to the number of populations identified by Citrus.