10:00a – 11:55p
(sailing canceled do to malfunction of committee boats)
coding matlab-storm
- RunDotFinder “skip existing” option on DaoSTORM I believe is interpreted as resume and in insightM as actually skip. This should be double checked and made clearer. Maybe we want a legitimate skip as well to be an option for DaoSTORM.
- Fixed up alternate binname functions (can pass it
'DAX'
to include the current daxfile name and append or prefix it with a new id. Can also pass it'#'
to insert image index (based on results of dir daxroot).
Mostly writing new multi-region version of ChromatinCropper
- Not all movies have bead data analyzed yet (I suppose I was planning to downsample all these files. I should probably set that up to run soon).
- note with no averaging and single round, daoSTORM seems to out perform insightM on bead-data
important observations in drift correction
- smoothing may work alright, but has serious issues at the boundaries. We use these boundary points to correct drift between movies, so this is an issue.
- we could compute a warp based on the bead images.
- fixed NaNs is
fastsmooth.m
(rather than usinginpaint_nans.m
, both from Matlab file server). Original author had tried to make it NaN friendly but got foiled by the+
and-
operators which still return NaN. (usenansum([a, -b])
rather thana - b
to subtract and ignore NaNs). - also switched to narrowed averaging window rather than just zero padding of ends during smoothing.
- still need to do something about vibrating beads.
STORM analysis
- Bead data from F12G1 not analyzed. Processing now on Cajal
- no data from G1G2 analyzed yet. Processing now on Cajal
- G1G2 data not nearly as bright as other data sets. Should collect more data with a fresh buffer when we get 750 time again.
- G1G2 slide is pretty low density on beads — lots of images have none.