WE FOUND THE STRAINS!!! 2017

This paper answers one of the most critical questions driving my PhD work, “Do microbial strains found on hospital surfaces colonize infants?” In short, yes but observing room to infant transfers is extremely difficult. The difficulty comes from 1) the hospital is an extremely low biomass environment, making it difficult to purify enough DNA to sequence metagenomes and 2) determining what constitutes a “strain” is somewhat debated and depends on your biological question.

Check out the paper for final details, but below are a few “behind the scenes” topics that didn’t make it into the paper.

1)    Upon attempting to write this paper, we were almost finished creating 16S rRNA amplicon libraries for a parallel project. These samples were paired with ddPCR for quantification of sample biomass. Below is a slide from one of my annual PhD dissertation committee meetings on August 25, 2015. This slide shows a few of the assumptions that formed the strategy for how many samples we aimed to pool in this paper. As of writing this copy (May 10, 2020), I wanted to publish this info somewhere, in case it’s helpful for others and before I forget these details (!!). At the time, UC Berkeley’s QB3 Core facility was requesting 50 ng of gDNA to create Illumina libraries using the Wafergen PrepX DNA library prep kit. While 50 ng was the requested minimum, we had fairly high success rates at getting decent metagenomic data from 1 ng of starting material. Using the ddPCR data we had at the time, that meant we needed roughly 19 samples per pool to generate usable metagenomic data for genome recovery.

2)    Clarifying what constitutes a strain is beyond the scope of this post! One of my coauthors has subsequently done fantastic work creating tools that help address this question (see his recent InStrain paper here) and has written follow-up studies using subsets of this data. Below, I show a strain story that didn’t make it into the paper, but was presented publicly at a presentation I gave and is available here. Infant 6 contains two strains of Klebsiella pneumonia. Infant 6’s Strain 1 was also recovered from Infant 5’s Sink samples. When this strain is in the sink, its genome replication rate (iRep) is ~1.45 and when in the gut the rate is 1.3. Strain 2’s iRep rate in the gut is 1.5 and this strain outcompetes the other. After some brief comparative genomics analyses we saw the sink adapted Strain 1 had differences that may have enabled it to thrive in the room environment (e.g. a suite of starvation associated genes, efflux pumps presumably to deal with hospital cleaners, etc.). This highlights the importance of resolving genomes to the strain level and observing these strains in various environments.