Percent Presence

[chasemc]

Percent Presence

I’ve had a lot of questions over the past couple of years about peak “percent presence”.

But, before I get into that, this isn’t a concept I came up with, or even implemented. The underlying code wraps a function filterPeaks() from the {MALDIquant} R package.

You find out more about MALDIquant from GitHub or the Strimmer Lab website.

---

The context is the following:

Lab: “Chase, I ran 3 replicates and set my percent presence filter to 50%. If 2 of 3 replicates have a mass peak at 400 m/z… Will that peak be retained?”

Chase[Panicking]: “I don’t remember, let me look at the code”

---

So I’m going to walk you through, in reproducible code, what exactly percent presence will remove in this scenario.

Install and load the MALDIquant package

install.packages("MALDIquant", repos = "http://cran.us.r-project.org")

library("MALDIquant")

The below function creates three fake mass peak lists (imagine three, peak-picked mass spectra). The arguments masses_1, masses_2, masses_3 can be changed so you can play around with scenarios. Intensity values are not important for this exercise and so are defaulted to a value of one.

create_peak_lists <- function(masses_1,
                              masses_2, 
                              masses_3){
  
  mass_spectrum_1 <- createMassPeaks(mass = masses_1,
                                     intensity = rep(1,length(masses_1)))
  
  mass_spectrum_2 <- createMassPeaks(mass = masses_2,
                                     intensity = rep(1,length(masses_2)))
  
  mass_spectrum_3 <- createMassPeaks(mass = masses_3,
                                     intensity = rep(1,length(masses_3)))
  
  list(mass_spectrum_1,
       mass_spectrum_2,
       mass_spectrum_3)
  
}

Create the mass peak lists as described above. Pay attention to the masses in each mass list (e.g. mass list 1 masses_1 has m/z values of 1, 2 and 3; mass list 3 masses_3 has m/z values of 1, 3 and 4)

mass_peak_lists <- create_peak_lists(masses_1 = c(1, 2, 3    ),
                                     masses_2 = c(1, 2, 3    ),
                                     masses_3 = c(1,    3, 4 )
                                     )

Here we loop over 0% to 100%, by intervals of 10%, calculating which peaks remain

percents <- seq(0, 1, .1)

res <- lapply(percents, 
       function(x){
         temp = MALDIquant::filterPeaks(l = mass_peak_lists, minFrequency = x)
         temp = MALDIquant::mergeMassPeaks(l = temp, method = "mean")
         temp@mass
       }
)

names(res) <- paste0(percents * 100, "%")

print(res)

Results are below, in the form
## percent presence
## m/z retained

Remember, it's the percent of samples a *m/z* must be found in in order to be retained.

## $`0%`
## [1] 1 2 3 4
## 
## $`10%`
## [1] 1 2 3 4
## 
## $`20%`
## [1] 1 2 3 4
## 
## $`30%`
## [1] 1 2 3 4
## 
## $`40%`
## [1] 1 2 3
## 
## $`50%`
## [1] 1 2 3
## 
## $`60%`
## [1] 1 2 3
## 
## $`70%`
## [1] 1 3
## 
## $`80%`
## [1] 1 3
## 
## $`90%`
## [1] 1 3
## 
## $`100%`
## [1] 1 3