P281424
/
AMR
mirror of https://github.com/msberends/AMR.git
1
0
Fork 0
Code Releases Activity

benchmarks on readme and speed improvement for direct mo codes

This commit is contained in:
dr. M.S. (Matthijs) Berends 2018-10-01 16:26:02 +02:00
parent ed17db0263
commit 6d761436f7
2 changed files with 123 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
R/mo.R
View File

@ -192,6 +192,13 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, allow_uncertain =
if (all(x %in% AMR::microorganisms[, property])) {
# already existing mo
} else if (all(x %in% AMR::microorganisms[, "mo"])) {
# existing mo codes
suppressWarnings(
x <- data.frame(mo = x, stringsAsFactors = FALSE) %>%
left_join(AMR::microorganisms, by = "mo") %>%
pull(property)
)
} else if (!is.null(reference_df)
& all(x %in% reference_df[, 1])
& all(reference_df[, 2] %in% AMR::microorganisms$mo)) {
@ -653,7 +660,7 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, allow_uncertain =
class(x) <- "mo"
attr(x, 'package') <- 'AMR'
attr(x, 'ITIS') <- TRUE
} else if (property %in% c("tsn", "year")) {
} else if (property == "tsn") {
x <- as.integer(x)
}

115
README.md
View File

@ -36,6 +36,7 @@ Erwin E.A. Hassing<sup>2</sup>,
* [Other (microbial) epidemiological functions](#other-microbial-epidemiological-functions)
* [Frequency tables](#frequency-tables)
* [Data sets included in package](#data-sets-included-in-package)
* [Benchmarks](#benchmarks)
* [Copyright](#copyright)
## Why this package?
@ -404,6 +405,120 @@ septic_patients # A tibble: 2,000 x 49
antibiotics # A tibble: 423 x 18
```
## Benchmarks
One of the most important features of this package is the complete microbial taxonomic database, supplied by ITIS (https://www.itis.gov). We created a function `as.mo` that transforms any user input value to a valid microbial ID by using AI (Artificial Intelligence) and based on the taxonomic tree of ITIS.
Using the `microbenchmark` package, we can review the calculation performance of this function.
```r
library(microbenchmark)
```
In the next test, we try to 'coerce' different input values for *Staphylococcus aureus*. The actual result is the same every time: it returns its MO code `B_STAPHY_AUR` (*B* stands for *Bacteria*, the taxonomic kingdom).
But the calculation time differs a lot. Here, the AI effect can be reviewed best:
```r
microbenchmark(A = as.mo("stau"),
B = as.mo("staaur"),
C = as.mo("S. aureus"),
D = as.mo("S. aureus"),
E = as.mo("STAAUR"),
F = as.mo("Staphylococcus aureus"),
times = 10,
unit = "ms")
# Unit: milliseconds
# expr min lq mean median uq max neval
# A 36.05088 36.14782 36.65635 36.24466 36.43075 39.78544 10
# B 16.43575 16.46885 16.67816 16.66053 16.84858 16.95299 10
# C 14.44150 14.52182 16.81197 14.59173 14.67854 36.75244 10
# D 14.49765 14.58153 16.71666 14.59414 14.61094 35.50731 10
# E 14.45212 14.75146 14.82033 14.85559 14.96433 15.03438 10
# F 10.69445 10.73852 10.80334 10.79596 10.86856 10.97465 10
```
The more an input value resembles a full name, the faster the result will be found. In the table above, all measurements are in milliseconds, tested on a quite regular Linux server from 2007 with 2 GB RAM. A value of 10.8 milliseconds means it can roughly determine 93 different input values per second. It case of 36.2 milliseconds, this is only 28 input values per second.
To improve speed, the `as.mo` function also takes into account the prevalence of human pathogenic microorganisms. The downside is of course that less prevalent microorganisms will be determined far less faster. See this example for the ID of *Burkholderia nodosa* (`B_BRKHL_NOD`):
```r
microbenchmark(B = as.mo("burnod"),
C = as.mo("B. nodosa"),
D = as.mo("B. nodosa"),
E = as.mo("BURNOD"),
F = as.mo("Burkholderia nodosa"),
times = 10,
unit = "ms")
# Unit: milliseconds
# expr min lq mean median uq max neval
# B 175.9446 176.80440 179.18240 177.00131 177.62021 198.86286 10
# C 88.1902 88.57705 89.08851 88.84293 89.15498 91.76621 10
# D 110.2641 110.67497 113.66290 111.20534 111.80744 134.44699 10
# E 175.0728 177.04235 207.83542 190.38109 200.33448 388.12177 10
# F 45.0778 45.31617 52.72430 45.62962 67.85262 70.42250 10
```
(Note: `A` is missing here, because `as.mo("buno")` returns `F_BUELL_NOT`: the ID of the fungus *Buellia notabilis*)
That takes up to 12 times as much time! A value of 190.4 milliseconds means it can only determine 5 different input values per second. We can conclude that looking up arbitrary codes of less prevalent microorganisms is the worst way to go, in terms of calculation performance.
To relieve this pitfall and further improve performance, two important calculations take almost no time at all: **repetive results** and **already precalculated results**.
Let's set up 25,000 entries of `"Staphylococcus aureus"` and check its speed:
```r
repetive_results <- rep("Staphylococcus aureus", 25000)
microbenchmark(A = as.mo(repetive_results),
times = 10,
unit = "ms")
# Unit: milliseconds
# expr min lq mean median uq max neval
# A 14.61282 14.6372 14.70817 14.72597 14.76124 14.78498 1
```
So transforming 25,000 times (!) `"Staphylococcus aureus"` only takes 4 ms (0.004 seconds) more than transforming it once. You only lose time on your unique input values.
What about precalculated results? This package also contains helper functions for specific microbial properties, for example `mo_fullname`. It returns the full microbial name (genus, species and possibly subspecies) and uses `as.mo` internally. If the input is however an already precalculated result, it almost doesn't take any time at all (see 'C' below):
```r
microbenchmark(A = mo_fullname("B_STPHY_AUR"),
B = mo_fullname("S. aureus"),
C = mo_fullname("Staphylococcus aureus"),
times = 10,
unit = "ms")
# Unit: milliseconds
# expr min lq mean median uq max neval
# A 13.548652 13.74588 13.8052969 13.813594 13.881165 14.090969 10
# B 15.079781 15.16785 15.3835842 15.374477 15.395115 16.072995 10
# C 0.171182 0.185639 0.2306307 0.2034135 0.224610 0.492312 10
```
So going from `mo_fullname("Staphylococcus aureus")` to `"Staphylococcus aureus"` takes 0.0002 seconds - it doesn't even start calculating *if the result would be the same as the expected resulting value*. That goes for all helper functions:
```r
microbenchmark(A = mo_species("aureus"),
B = mo_genus("Staphylococcus"),
C = mo_fullname("Staphylococcus aureus"),
D = mo_family("Staphylococcaceae"),
E = mo_order("Bacillales"),
F = mo_class("Bacilli"),
G = mo_phylum("Firmicutes"),
H = mo_subkingdom("Posibacteria"),
times = 10,
unit = "ms")
# Unit: milliseconds
# expr min lq mean median uq max neval
# A 0.145270 0.158750 0.1908419 0.1693655 0.218255 0.300528 10
# B 0.182985 0.184522 0.2025408 0.1970235 0.209944 0.243328 10
# C 0.176280 0.201632 0.2618147 0.2303025 0.339499 0.388249 10
# D 0.136890 0.139054 0.1552231 0.1518010 0.168738 0.193042 10
# E 0.100921 0.116496 0.1321823 0.1222930 0.129976 0.230477 10
# F 0.103017 0.110281 0.1214480 0.1199880 0.124319 0.147506 10
# G 0.099246 0.110280 0.1195553 0.1188705 0.125436 0.149741 10
# H 0.114331 0.117264 0.1249819 0.1220830 0.129557 0.143385 10
```
Of course, when running `mo_phylum("Firmicutes")` the function has zero knowledge about the actual microorganism, namely *S. aureus*. But since the result would be `"Firmicutes"` too, there is no point in calculating the result. And since this package 'knows' all phyla of all known microorganisms (according to ITIS), it can just return the initial value immediately.
## Copyright
[![License](https://img.shields.io/github/license/msberends/AMR.svg?colorB=3679BC)](https://github.com/msberends/AMR/blob/master/LICENSE)