In Sorbus-torminalis/twopointzerothree: Detects Duplicate Sequences of Offset Values

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

twopointzerothree

twopointzerothree is motivated by an expression of concern concerning a paper coauthored by Dr Pruitt following the discovery, amongst other issues, that

Seven consecutive values from Theridon agressiveness assay #1 differ exactly by + 2.03 from another exact sequence in aggressiveness assay #1 in a different treatment group

I wanted to know how this could be done. As far as I know, standard tools (n-grams) only detect duplicate sequences, not duplicate sequences with an offset.

twopointzerothree develops a method for detecting duplicate offset sequences by searching for repeat sequences of perfectly correlated numbers. These repeat sequences are either duplicates or offset duplicates.

twopointzerothree can be installed with

remotes::install_github("Sorbus-torminalis/twopointzerothree.git")

twopointzerothree requires a vector of numbers as input.

library(twopointzerothree)
data("kp2014")# part of the data from Keiser & Pruitt 2014
sequence_find_all(vec = kp2014$`Theridion murarium Aggressiveness...4`)

The result give the position and values of the first and second sets of duplicated values and any offset.

If data in several columns need to be checked, they should be appended into one vector.

sequence_find_all(
  vec = c(kp2014$`Theridion murarium Aggressiveness...4`,
          kp2014$`Theridion murarium Aggressiveness...5`,
          kp2014$`Theridion murarium Aggressiveness...6`))

To see the duplicate sequences in situ, use sequence_show.

sequence_show(kp2014, meta_cols = 1:3, test_cols = starts_with("Theridion"))

Notes

twopointzerothree will report the all the duplicate sequences it can find.

twopointzerothree will not report sequences of identical values as duplicates as cor(rep(1, 3), rep(2, 3)) returns NA.

If very short sequences (e.g. length 3) are searched for, there is a high risk of false positives. These can be identified by having offset values that are not consistent.

twopointzerothree reports duplicate values. It does not report why.

If you find this package useful or have ideas about how it could be improve, please leave a message in the issues tab.

Sorbus-torminalis/twopointzerothree documentation built on Aug. 18, 2022, 2:30 a.m.