11:00 am – 5:00 pm, 10:00 pm – 11:55 pm
- break down STORM run
- # transfer data
- # fit molecule localizations
- finish compiling oligo-array output folders.
- finish writing review
- first complete draft finished.
- just need to read through this again from top to bottom and should be ready to go.
- revised and submitted
- work on slides for meeting with Ting group on Tuesday
- forgot to take bead images ! Darn it, I think this kills my last rather nice multi-color black data set.
Tuck doesn’t respond to process kill commands (at least not for many processes). (no idea why). window claims that kill command executed successfully, but task manager still reports that jobs are running. I suppose I should kill them in task manager directly. This is probably why Tuck lost so badly to Cajal in the probe design race (30K to 6K)
Control sequences still not the right length!
Perhaps I overwrote the 9-5 probes with the 9-4 probes run still with the wrong parameters (why 9-4 were wrong parameters I have no idea). Looking in the folder though the 9-4 probes are indeed the wrong length, and my collected list of probes contains the 9-4 versions. 9-5 and 9-6 folders are now empty — I guess they ‘moved’ rather than ‘copied’…
- synced up with matlab-storm and matlab-functions with release / jeff
- still need to push my matlab-storm changes back up.
- Jeff’s matlab-storm updates broke STORMfinder find dots option.
10:00 am – 12:00 am
Update on running Analysis
- Cajal and Tuck still building library, 20-40% complete.
- Won’t be done for another 24 hours at the earliest.
- maybe longer as the day brings competition of these cores and my programs are “polite” and launch at low priority.
- Interaction analysis still running, should finish before tonight
- refitting BB black D12
Internal black analysis
- D11 — conventional images useless, STORM images not promising. fortunately I already repeated this one myself
- D12 — out of focus spots are overly large
- spots also pretty dim for Alexa647
- large spots only get off-center localizations when multiple spots are called in the same frame.
- looks promising if we can refit the data.
- first pass at refitting totally mismatched files / images. Copying data, will repeat this.
- D08 — D08_08_05_2015 conventional images are more or less blank. Most STORM images very close to blank.
- D09 — data looks reasonable. Reprocessed.
- D09-10 conv looks okay, lots of storm movies missing or no reasonable data. 2-10 might be good.
- fiducial drift correction failed or extremely poor.
- Correlation based drift correction looks to be sufficient accuracy
- D10 S1-647, D09 S3-750
- D11 S1-647, D12 S3-750
- D08 S1-647
- D10 S1-647
- E05 S1-647 + E04 S3-750
- E08 S1-647 + E07 S3-750
passaged cells, replaced media.
tossed old cells (after 2 months in culture).
- E05 S1-647 + E04 S3-750
- F04 S1-647 + F03 S3-750
- L4E2 S2-647, L4E1 S3-750
Chromatin region analysis
- processing L4E17 data (15 kb of BXC)
- processing L4E17-19: 45 kb of BXC
- Need to fix G09 is a (75%) internal part of C05, not an end to end domain
- processed black 2 color data: L4E1 + L4E2.
- 750 missing in most images,
- partly due to laser off-center
- largely due to buffer death
- re-processed internal black domains as indicated above.
- some results
Attempting batch jobs on Odyssey
- see Bogdan’s email
- array launch still goes into queue
- jobs actually launched.
- individual output files and error reports also generated.
- what if we request 500 cores and then do a bunch of
system([task,' &']) commands in a loop? On a desktop these go out to separate cores. Lets see if this simple approach can work on the remote.
- all java commands die with memory errors. (didn’t actually expect them all to launch, so maybe this is progress).
- actually started running all these java files. I see even _oligos.txt files for a bunch of genes, some even with data and probes.