In PaolaCognigni/CTraxHelper: CTrax Helper

Track analysis workflow:

Track input

Select and upload CTrax files into R objects

to do: define a trx class for the data frame and some wrapper class for the experiment list

Script 0.1:

input dir and file information therein
report file list and name list
output raw list

Select and binarise the light protocol from a csv file

to do: automate threshold

Script 0.2:

input light protocol file and protocol structure info
report constructed time protocol (vs raw data)
output t time protocol data frame

Data processing 1 - global

Garnish data combining time protocol information

to do: impose correct var type (factor, int, num) to variables

to maybe do: add some vars (log transforms and abs), think of more?

Script 1.1:

input raw and t
report: not sure, as this is a behemoth of data. Could do the track integrity and a general track plot by id~genotype
output garn list

Data processing 2 - quadrant protocol

Calculate PI (by fly and by repeat)

Script 2.1:

input garn
report boxplots of PIs by genotype
output PIfly and PIrept (PIbyfly and PIbyrept too?)

Data processing 3 - section3 protocol

Tidy section 3 time protocol

generalise to a given known protocol (frames on, frames repeat)

add drift?

Script 3.1: (note: there is no reason not to run this immediately after 0)

input t and protocol parameters
report maybe time offsets and such
output tsynth data frame and perhaps the progress time annotator

Extract section 3 to produce a subsetted data frame of all genotypes

to do: once the exp class is defined, create a listtodf method

to do: think more seriously about the structure of tsynth (esp the strange chunk 3 declaration)

Script 3.2:

input raw or garn and tsynth
report something like mean trends of various variables over the protocol? or a summary of data (which vars, how many ids per genotype). or the non-binned data could be generated here (as in var paths over time, overplotted by id and blink), or that could be a separate report or script.
output s3df

Collapse stats to 1s bins

to do: look at median and standard deviation too

Script 3.3

input s3df
report boxplots of meaningful variables by second and repeat
output s3mean (and possibly s3median and s2df)

Figure design

try to work out what this is good for, if anything

show plot of PCA for a laugh

Script 3.4

input s3df
report path patterns by second and repeat
output s3figs

Data processing 4: Virtual quadrant design

Generate synthetic fly paths by fragmenting and collating on and off real paths

idea: define x and y in radial coords from the centre of the plate (somewhere in garn there may still be an arena list, or I can pull out its definition) to simplify track assembly by rotating things about instead of shifting

to do: decide/try either chunking by classifier or going entirely blindly.