adsteen / enzalyze Goto Github PK

Right now parse_numeric parses a character vector as a number and a unit. However, it can currently fail in any of many ways - for instance, if different units are specified down the column (currently it picks the first one and ignores the rest) or if any element in the vector gets spit into more than 2 strings.

d$Rtime <- ymd_hm(paste(the.date, d$time))

this code parses data through lubridate of the format; _hm

need a try catch to account for data potentially parsed as _hms

As written, safe_NLS() will never work with any exponential formula that doesn't include A and time as variables. Currently A and time are hard-coded into generate_exp_guess(); should be passed as parameters.

d[ , "RFU"] <- as.numeric(gsub(",", "", d$RFU))

convert d[ , "RFU"] so that it is flexible for any fluorescence variable; not just "RFU"

d$"RFU" <- as.numeric(gsub(",", "", d$"RFU"))

need to find a way to convert numeric values for that column if it's not called "RFU"

The generate_exp_guess() uses to create a linear model is lm(unlist(select(df, ln.ycol)) ~ unlist(select(df, !!xcol))). I think this is relatively safe, but there is surely a more elegant and easier-to-read way to do it.

first few steps of function need to not be hardcoded in as substrate or RFU