FaRMI

I will add now a special collection of farming studies here as many of them are just candidates for the Ig nobel prize.

The most recent study introduces FaRMI, a “bacterial relative abundance farm home microbiota index”, probably introduced as the authors couldn’t find anything else. It reminds me very much to the polygenic risk score that rescues your study if you could not find the gene.

Asthma prevalence has increased in epidemic proportions with urbanization

Clearly already the first sentence is wrong if we look at the following plot.

Urbanization happened in the late späten 19th century and not after 1950. Source of plot: The prevalence of asthma in children: a reversing trend ( ERJ 2005 )

Unfortunately, the difference between farm and non farm children is never explained in the Kirjavainen et al. paper.

What is for example the average distance of a non farm house to a farm house? Are there any joint school or sports activities of children from farms and non farms (allergens travel in the classroom)? And why is there such a strong conclusion in the title?

Farm-like indoor microbiota in non-farm homes protects children from asthma development

A lower risk score is not equivalent to protection.

And did somebody of the authors or reviewers ever look at the plots?

I do not understand Figure 3c. It even makes only sense when I cross out the top labels. But even then it menas: GABRIELA does not show a significant replication.

or tables?

What should these values tell us? 5 times p<0.05 in a grid of 6×11=66 tests? Quantile regression that has been “adjusted” by the data?

And isn’t that  just an association that may have a rather simple explanation?

As FaRMI is weakly associated with muramic acid concentration in dust, the authors make Gram-positive bacteria responsible for the effect. The rhizosphere of soil is extremly rich of bacteria. The world’s first soil atlas showed  hundreds of taxa but never differentiated between water resistant, gram positive and less water resistant gram-negative taxa. Maybe Gram positive Streptococcaceae are ubiquitous and depend on where you draw your samples?

FaRMI is found in non farm / rural children by bacterial/archaeal operational taxonomic units (OTUs) of soil origin which basically confirms my initial assumption: There was the same contamination of soil both in farm and non-farm homes if we look at supplement table 6 where walking indoors with outdoor shoes results in significant higher FaRMI values…

In conclusion, while the asthma-protective effect of farming is intriguing, it has little practical relevance unless the protective effect can be functionally transferred to non-farming environments.

I do not find this data derived score intriguing. Maybe the microbiome hype is already over.

Our results warrant translational studies to confirm the causal relationship through indoor microbial exposure-modifying intervention that may also form a novel strategy for primary asthma prevention.

Good luck with your translational studies, as we are now somewhere in the nowhere.

 

BTW – The scripts at Github are useless references to shell and Python scripts that will never run due to “—” characters. And what about that baby girl code?

writerow <- paste("Eigenvalue min / max: ", min.eigen, " / ", max.eigen, sep="")
write(writerow, file=eigenfile, append=F)
writerow <- paste("Sum of all eigenvalues: ", round(neg.eigensum, digits=6), sep="")
write(writerow, file=eigenfile, append=T)
writerow <- paste("Sum of all eigenvalues (negatives as 0): ", round(nonneg.eigensum, digits=6), sep="")
write(writerow, file=eigenfile, append=T)
writerow <- "Eigenvalues (pos & neg): "
write(writerow, file=eigenfile, append=T)
writerow <- paste(pcoa$value$Eigenvalues, collapse="\t")
write(writerow, file=eigenfile, append=T)
writerow <- "Percents (Negatives as negatives): "
write(writerow, file=eigenfile, append=T)
writerow <- paste(paste(neg.percent, " %", sep=""), collapse="\t")
write(writerow, file=eigenfile, append=T)
writerow <- "Percents (Negatives as 0): "
write(writerow, file=eigenfile, append=T)
writerow <- paste(paste(nonneg.percent, " %", sep=""), collapse="\t")
write(writerow, file=eigenfile, append=T)

Using R heredoc syntax I can rewrite 18 unreadable to 10 readable lines. And 9x disc access to 1x just doing

tmp <- 'Eigenvalue min / max: min.eigen / max.eigen
Sum of all eigenvalues: neg.eigensum
Sum of all eigenvalues (negatives as 0): nonneg.eigensum
Eigenvalues (pos & neg): pcoa
Percents (Negatives as negatives):  neg.percent %
Percents (Negatives as 0): nonneg.percent %'
for (i in c("min.eigen","max.eigen","neg.eigensum","nonneg.eigensum","pcoa$value$Eigenvalues","neg.percent","nonneg.percent") ) { tmp <- gsub(i,get(i),tmp) }
write(tmp, file=eigenfile)

And why moving to SAS for a simple logistic regression? Is there anyone else in the academic world who pays $8,700 annually for a basic SAS Windows Analytics package?