-omics in 2013

Just how many (bad) -omics are there anyway? Let’s find out.

1. Get the raw data

It would be nice if we could search PubMed for titles containing all -omics:

*omics[TITL]

However, we cannot since leading wildcards don’t work in PubMed search. So let’s just grab all articles from 2013:

2013[PDAT]

and save them in a format which includes titles. I went with “Send to…File”, “Format…CSV”, which returns 575 068 records in pubmed_result.csv, around 227 MB in size.

2. Extract the -omics
Titles are in column 1 and we only want the -omics, so:

cut -f1 -d "," pubmed_result.csv | grep -i omics > omics.txt
wc -l omics.txt
# 1770 omics.txt

3. Clean, rinse, repeat…
We want just a list of -omics words. Time to break out the R. After much trial and error, I ended up with this. Ugly and far from optimized, but it (mostly) works. I say mostly, because I know of at least one case which is not detected: stain-omics.

library(stringr)

omics <- readLines("omics.txt")
omics <- strsplit(omics, " ")            # split titles on space
omics <- unlist(omics)                   # convert to vector of words
omics <- omics[grep("omics", omics)]     # just the -omics words
omics <- gsub("[\"\'\\.:\\?\\[\\]]", "", omics, perl = T)  # remove symbols, punctuation
omics <- tolower(omics)

m <- data.frame(a = omics, b = str_match(omics, "^(.*?omics)-")[, 2])  # matches e.g. "genomics-based"
omics <- ifelse(is.na(m$b), as.character(m$a), as.character(m$b))                       

m <- data.frame(a = omics, b = str_match(omics, "-{1,}(.*?omics)$")[, 2])  # matches e.g. "phospho-proteomics"
omics <- ifelse(is.na(m$b), as.character(m$a), as.character(m$b))

omics <- unlist(strsplit(omics, "\\/"))  # split e.g. "genomics/proteomics"
omics <- omics[grep("omics", omics)]     # just the -omics words again

# OK we're down to the edge cases now :)
omics <- gsub("applications", "", omics)
omics <- gsub("\\(meta\\)", "meta", omics)

4. Visualize
The top 20 -omics in 2013 and the less popular:

omics.freq <- as.data.frame(table(omics))
omics.freq <- omics.freq[ order(omics.freq$Freq, decreasing = T),]
ggplot(head(omics.freq, 20)) + geom_bar(aes(omics, Freq), stat = "identity", fill = "darkblue")
                             + coord_flip() + theme_bw()
# and the less popular
subset(omics.freq, Freq == 1)

On the right, the top 20. Click for a larger version of the graphic. Top of the list so far for 2013 is proteomics, followed by genomics and metabolomics. Listed below, those -omics found only once in titles from 2013. Some shockers, I think you’ll agree (paging Jonathan Eisen). omics Freq aquaphotomics 1 biointeractomics 1 calciomics 1 cholanomics 1 cytogenomics 1 cytokinomics 1 econogenomics 1 glcnacomics 1 glycosaminoglycanomics 1 interactomics 1 ionomics 1 macroeconomics 1 materiomics 1 metalloproteomics 1 metaomics 1 metaproteogenomics 1 microbiomics 1 microeconomics 1 microgenomics 1 microproteomics 1 miromics 1 mitoproteomics 1 mobilomics 1 morphomics 1 museomics 1 neuromics 1 neuropeptidomics 1 nitroproteomics 1 nutrimetabonomics 1 oncogenomics 1 orthoproteomics 1 pangenomics 1 petroleomics 1 pharmacometabolomics 1 pharmacoproteomics 1 phylotranscriptomics 1 phytomics 1 postgenomics 1 pyteomics 1 radiogenomics 1 rehabilomics 1 retrophylogenomics 1 romics 1 secretomics 1 sensomics 1 speleogenomics 1 surfaceomics 1 surfomics 1 toxicometabolomics 1 vaccinomics 1 variomics 1

Top 20 -omics in PubMed titles, 2013

Never heard of romics? That’s OK. It’s a surname.