@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra 1/n Thanks Pat. Happy to discuss, Varun. In the meantime, here's a quick (incomplete) recommended reading list on handling the "noise" in low biomass microbiome studies:
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra 2/n "Reagent and laboratory contamination can critically impact sequence-based microbiome analyses" (The seminal "Salter paper.") @Zannah_Du bmcbiol.biomedcentral.com/articles/10.11…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du 3/n "Recognizing the reagent microbiome" nature.com/articles/s4156…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du 4/n "How low can we go? The implications of low bacterial load in respiratory microbiota studies" pneumonia.biomedcentral.com/articles/10.11…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du 5/n "Garbage in, garbage out: Wrestling with contamination in microbial sequencing projects" - Fantastic blog post on the topic. I agree w/ every word.@NoahFierer fiererlab.org/2018/08/15/gar…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 6/n We haven't written up our approach, but you can see how we handle it in this paper, which I summarized in a twitter walkthrough. atsjournals.org/doi/abs/10.116…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 7/n Basically: 1) sequence a ton of negative controls, representing any potential source of contamination, 2) transparently report the taxa in your negative controls, 3) compare the taxa in your negative controls with those of your low-biomass specimens...
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 8/n 4) Strongly consider using absolute quantification in addition to relative quantification of taxa... 
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 9/n 5) Exclude entire specimen types (not individual specimens!) when they are collectively indistinguishable from background - e.g. the nasal rinse specimens in this experiment.
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 10/n 6) We do NOT recommend "subtraction" of suspect taxa just because they're ID'd in negative controls. See discussion in above links; this is likely substituting one bias for another. Also, overlap between environmental taxa and specimen taxa may --> false negatives.
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 11/n (Not to mention the issue of index switching/barcode error, which is an underappreciated source of false signal in your negatives. Subtracting 'negative' taxa will remove legitimate signal.) biorxiv.org/content/early/…
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 12/n 7) Treat your negatives as a potential "source community" for taxa of interest and report them accordingly/transparently (e.g. the attached figure from our article above).
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 13/n 8) Be honest w/ yourself and readers about biologic/microbiologic plausibility for the taxa you're finding.
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 14/n 9) Ask if your microbial signal correlates w/ non-microbiologic indices that matter. E.g if a taxon is inversely correlated with total bacterial burden, suspect contamination. If it correlates positively w/ host inflammation, that's (soft, indirect) support for its reality.
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 15/n 10) Use non-sequencing techniques to confirm any important taxa of interest. Cultivation, targeted PCR, transcriptomics, etc..
@PatSchloss @vnsriniv @watermicrobe @gregcaporaso @DNAkendra @Zannah_Du @NoahFierer 16/16 I'll stop there... lots more in those above articles/blog post. There's no one settled approach, and reasonable people may disagree about details, but there's really no debating that in 2018, low-biomass microbiome studies need to take the above issues into consideration.
