9:55a – 10:00p

# Chromatin Project

## cell fixing and staining

- fix new Kc cells
- prep for in situ
- treat with RNase
- stain new kc cells with F03F04 and G01G02

# Ph

## Data analysis

- wt data has half the number of frames (54,000 vs 102,000) and half the average number of localizations
- this creates a problem:
- Fewer localizations means that increasing the binSize in the clustering algorithm causes visibly apparent clusters to be split apart.
- More total localizations increases the fraction of small clusters weakly connected by chance localizations, artificially increasing the cluster size.

- potential solution
- for a fixed number of clusters, if we double the number of total localizations, we should be able to sample the area at twice the coverage and still maintain the same localization density. This should also avoid spurious new localization linking existing clusters (which would happen if we didn’t change the sampling density). To halve the area (twice the coverage) we change the sampling dimension by sqrt(2). So lets rescale the binsize by the root of the localizations (or localizations per cell).