11:00 A – 7:30P, 8:45P – 10:00P
Cell staining
- label with m-HP1a -405,750 and rb-H3K27me3 -488 (top) or rb-H3K27ac -488 (bottom row)
Polytene cell staining
- label with m-HP1a-Cy3b and rb H3K4me3 -488 or rb H3K27me3-488.
- check staining on confocal
- some bacteria now detectable on these sections
- mount in prolong gold, let set O/N.
Confocal
Pc
Coding /Data Analysis
- Exploring more methods to filter out non-switching dots
- filter_nonblinkers tries use knnsearch to find nearest neighbors as a clustering approach, and then use look at variance in distribution of frames found for dots in the cluster.
- I think the true clustering as in (Analyze_K27me3) might work better. Need to partition image into smaller sub-images first to keep the speed.
- Also needs a dataset with good conventional images. My previous investigations showed differences between conventional intensity and STORM dots robustly flags regions where weak non-blinking emitters give false signal.
- Analyzing Pc-GFP Pc data from 04/24/12 (slide 155, stain started 4/4/12).
- Making summary slide of cluster stats for Ajaz
- Added new manual cell boundary selection to findclusters_percell.m For moderate numbers of cells (~100) this is a bit tedious, but more productive than specifically tuning parameters to get only half decent segmentation. A fast 2D clustering algorithm might work alright combined with the appropriate smoothing filter…