Population-level analysis of gut microbiome...
Transcript of Population-level analysis of gut microbiome...
Population-level analysis of gut microbiome variation
Gwen FalonyDepartment of Microbiology and ImmunologyRega Institute, KU LeuvenVIB Center for the Biology of Disease
How human are you?
•There are 100.000 times more bacteria on your body than there are humans on earth.
•You have 10 times more bacterial cells than human ones.
•You have 100 times more bacterial genes than human ones.
The human gut microbiota•expands our digestive capacity,
•produces essential nutrients
•increases colonization resistance against pathogenic intruders
•assists in the detoxification of xenobiotics
Towards health monitoring through gut microbiota screening
A quest for causality
Turnbaugh et al., Nature 2006
Microbiota-induced obesity
• Gut microbiota contributes to obesity
• Obesity is characterized by enhanced energy harvest efficiency
• Increased relative abundance of Firmicutes is characteristic for an obesity-associated microbiota
A request for skepticism
Hanage et al., Nature 2014
Can experiments detect differences that matter?Does the study show causation or just correlation?What is the mechanism?How much do experiments reflect reality?Could anything else explain the results?
Microbiota modulation as therapy
Fecal Microbiota Transplantation
Vrieze et al., Gastroenterology 2012
Association ≠ Causality
Confounder analyses are
required
Flemish Gut Flora Project
www.facebook.com/VlaamsDarmfloraProject
www.vib.be/nl/mens-en-gezondheid/darmflora-project
Sample collectionCurrent status: 3400 samples collected
Falony*, Joossens*, Wang*, Vieira-Silva* et al., Science 2016
503 potential microbiome covariates analyzedMicrobiome analysis: 16S rRNA gene amplicon sequencing
Worldwide sampling 3,948 samples, 664 genera …
… and counting (est. 40,000 samples required to cover full diversity)
Enterotypes - microbiota configurations in gut microbiome space
Arumugam*, Raes* et al., Nature 2011
Microbiota constellations independent from gender, age, and nationality
Enterotype configurations in the Flemish population
Prevalent microbiome constellations differ in relative abundances of enterotype identifiers
Love thy neighborsCollaboration with Cisca Wijmenga and Jingyuan FuUniversity of Groningen, The Netherlands
Falony et al., Science 2016
Zhernakova et al., Science 2016
Microbiome covariate identification• Discovery: FGFP (N=1106)• Replication: LifeLines Deep (N=1135)• 69 FGFP covariates• 92% replication
The Flemish Gut Flora Project perspective
1106 samples503 metadata variables69 microbiome covariates18 non-redundant covariates
Stool consistency is associated with microbiota richness and enterotypes
Vandeputte et al., Gut 2015Falony et al., Science 2016
A forgotten covariate
Fruit consumption
Meat consumption
Bread type preference
Soy products/yoghurts
Go Belgium!
Beer consumptionDark chocolate preference
Diet matters!
But so does taste …
Disentangling disease/treatment signaturesIdentification of microbiome-drug associations
Drugs interacting with covariate-associations
Direct associationsantibiotics, laxatives, immunosuppressants, hormones
Falony et al., Science 2016
Forslund*, Hildebrand*, Nielsen*, Falony* et. al., Nature 2015
Many genera previously associated with disease are also linked to other microbiome covariates
Falony et al., Science 2016Faecalibacterium: UC, but also RBC, laxatives, and age
No significant association between early life events and adult microbiota composition
Falony et al., Science 2016
Cesarean vs vaginal deliveryBreastfed vs not breastfedHome vs hospital birthRural vs urban childhood
503 potential covariates – 7.7% non-redundant cumulative effect size
What did we miss?
Lima-Mendez*, Faust*, Henri* et al., Science 2015; Faust et al., Curr Opin Micr 2015
Tara Oceans
Environmental factors
explain <15% of variation
Species interactions, founder effects, dynamics - and currently unassessed covariates
Conclusions – the Flemish Gut Flora Project experience
M. JoossensS. Vieira-Silva
J. WangY. DarziK. Faust
A. KurilshikovM.J. Bonder
M. Valles-ColomerD. Vandeputte
R.Y. TitoS. Chaffron
L. RymenansC. VerspechtL. De Sutter
G. Lima-MendezK. D’hoe
K. JonckheereD. Homola
R. GarciaE.F. Tigchelaar
L. EeckhaudtJ. Fu
L. HenckaertsA. Zhernakova
C. WijmengaJ. Raes
• Replication is possible in microbiome research• Many covariates have been identified, more remain
unknown. And all of them are potential confounders • Effect sizes are smaller than generally assumed• Careful study design, adequate cohort sizes, and
investment in sampling logistics and quality control are crucial
Thank you for your attentionQuestions?