tdurham86
commented
9 years ago I took a crack at this one and simply binarized our expression data to show whether or not a gene is expressed. The threshold for expression is a fluorescence value >2000 from the minimum expression value for that gene and also >10% of the maximum expression for that gene (based on the gene expression calls in Murray, et al. 2012). We may still need to tweak this, or try other things, but this should be at least a pretty good place to start.
We need to make the gene expression comparable in some way between time points. We will NOT be able to compare expression values directly due to the way these measurements were taken. The magnitude of fluorescence observed in a given time series is dependent on a variety of factors that are totally unrelated to the actual level of gene expression, including the number of times the fluorescent construct was integrated into the genome, where in the genome these integrations took place, etc. Thus, we will be comparing expression patterns across time and lineage (where and when we see expression), not expression levels. Past analyses of these data have used a threshold of 10% of maximal fluorescence to call a gene as expressed, so perhaps as a first pass we can normalize our fluorescence values for each gene to the maximal fluorescence in that gene and call any gene "on" in a given cell if it has a normalized fluorescence value > 0.1 in that cell.