Challenges and limitations for improving feed efficiency from metagenomics data. Oscar González Recio, I. Guasch, A.

Transcription

1 Challenges and limitations for improving feed efficiency from metagenomics data Oscar González Recio, I. Guasch, A.

2

3 Efficiency Ability to accomplish a job with the least waste of time, resources and effort Residual Feed Intake (RFI) Ratio of energy produced by a machinery to the energy supplied to it FEratio=Milk/DMI

4 GENETIC PROGRESS litres/year 350 kg liveweight 8.6 litres/kg LW 9000 litres/year 600 kg liveweight 15 litres/kg LW Doubled the efficiency at feed utilization J. Pryce

5 Spanish champions 1990 and 2015 Increased benefit of 290 per lactation

6 Maintenance cost Size score BW EBV Milk EBV Maintenance cost González-Recio et al., (2007)

7 BW needs RFI Feed Saved Pryce et al., (2015)

8 Feed efficiency in Selection indices

9 Proxies Expensive to measure. Infraestructure Genomic Selection (gdmi) Proxies: LW, MIR, Methane, Microbiota Challenges from MICROBIOTA more than a proxy: holobiont organism with own genome Interacts with the host

10 2-12 % loss energy Latifa Ouatahar

11 Host genetic control

12

13 81 cows Rumen content I. Guasch, G. Elcoso Alex Bach

14

15 81 cows Rumen content I. Guasch, G. Elcoso Mothur: Classify reads in Operational Taxonomic Units (OTU) at the genus level Alex Bach

16 Core taxa composition

17

18 16S rrna association between genus composition and FE related traits

19 16S rrna

20 16S rrna Detect associations between FE and microbiota Linear regression, adjustment Benjamini-Hochberg (Limma package in R) Bayesian Zero inflated models (HPD 90-95%) (mixture models) Random Forest (Variable Importance)

21 16S rrna Detect associations between FE and microbiota Zero inflated models

22 16S rrna Detect associations between FE and microbiota Feed efficiency Random Forest Residual Feed Intake

23 RFI FE FE= ρ=-0.35 RFI=

24

25 16S rrna Variable selection Genus: Prevotella Lachnospira Coprococcus Shuttleworthia Ruminococcus Methanobrevibacter CF231 Unespecified genera from families: Paraprevotellaceae RF16 BS11 Christensenellaceae

26 16S rrna Variable selection 16S amplicons (selected taxa)

27 81 cows Rumen content 15 HIGH/15 LOW I. Guasch, G. Elcoso Alex Bach

28 Metagenome 31 samples from MiSeq Illumina

29 Metagenome Bioinformatics

30 Metagenome. Taxonomy High vs Low Acidobacteria environmental samples <Bacteria> FCB group Bacteroidetes Fibrobacteres Fusobacteria Proteobacteria Bacteria PVC group Chlamydiae Verrucomicrobia Spirochaetes Terrabacteria group Actinobacteria <phylum> Firmicutes Tenericutes unclassified Bacteria Bacteria candidate phyla candidate division SR1 Candidatus Saccharibacteria Euryarchaeota unclassified Bacteria (miscellaneous) Methanobacteriaceae Methanobrevibacter Methanosphaera cellular organisms Thermoplasmata root Babesia Entodiniomorphida Hymenostomatida Ophryoglenina Alveolata Ciliophora Intramacronucleata Oligohymenophorea Tetrahymenina Parameciidae Philasterida Sporadotrichida Stentoridae Eukaryota Entamoeba saccharomyceta leotiomyceta Fungi Saccharomycetales Opisthokonta Bilateria Neocallimastigaceae Anaeromyces Neocallimastix Teleostomi Beatriz Delgado Insecta Tritrichomonas foetus Oomycetes Euphyllophyta Viruses unclassified sequences

31 Metagenome. metagwas - FE 174,247 8,713

32 Metagenome. metagwas FE (448 contigs) Correlation analyses Cluster Analysis

33 Metagenome. metagwas FE (448 contigs) 16S amplicons (selected taxa) Metagenome (selected contigs)

34

35 Metagenome. metagwas - DMI Filtering with Information Gain Logistic regression Parity+DMIgroup+e

36 Metagenome. metagwas - DMI

37

38 Variance components estimation Fixed model FE i = μ + DIM ij + Parity ik + e i Alejandro Saborio microblup (16S) 2 FE i = μ + DIM ij + Parity ik + m im + e i m: N K σ metagblup (0, ) 16S 16S FE i = μ + DIM ij + Parity ik + m metaim + e i m 2 meta : N(0, Kmetaσ meta ) Bayesian regression BGLR package (de los Campos and Perez Rodriguez)

39 Variance components estimation

40 Variance components estimation

41 Take home message Rumen microbiota Feed Efficiency HIGH vs. LOW FE cows, (and DMI!) A relevant proportion of feed efficiency is explained by the microbiota. PROXIES Beware of amplicons! Association but not discriminate high vs low Larger sample size or WMG sequencing Limitations Find the added value as a proxy Develop better statistical models that account for the microbiome x host Validate the results

42 Acknowledgements Isabel Guasch Alex Bach Guillermo Elcoso Beatriz Delgado Carmen Gonzalez Latifa Ouatahar Alejandro Saborio Adrian López METALGEN IRTA BLANCA from the Pyrenees FISABIO MINECO

43 animal breeding

44 Microbiome 44 Perturbe microbiota G K G x K

45 KEGG Pathways for known significant genes

46 Thank you!! Any questions?

47 8173 contigs top 95% IG Post-GWAS analyses (cluster analyses) Validation

48 RFI and EB

49 CP=16% Nel=1.64 Mcal/kg NDF=27%

50 Results Qiime vs Mothur

51 How to perturbe the microbiota Breeding programs Economic value of microbiota vs Correlation with feed efficiency Option 1) RA of a core microbiome Option 2) Select for the effect of microbiome from mixed model as a pseudophenotype Biotechnology Pro/Prebiotics Vaccines Additives Gene-Edition

52 Main limitation of metagwas Ignores the complexity of the microbiota composition, analysing one microbe/contig at a time Complex interactions between RA of microbes/contigs Metagenome as a whole

53 In-farm sequencing Go moooo!

54

55 Results Mothur identified 72 OTUs, 14 phylum, 41 genera Analysed: Relative abundance >0.1% OTUs appearing in all animals Allowed only 1 missing 25 OTU remained

56 Variance components estimation model DIC Residual Variance GenomicVariance Genomic h2 Microbial h2 Fixed Model NA NA NA GBLUP NA mblup NA NA 0.01 metablup NA NA 0.09 GBLUP+BL NA GBLUP+mBLUP GBLUP+metaBLUP model Cor(y,GEBV) Cor(y,Micro) Cor(y,y_hat) Fixed Model NA NA 0.54 GBLUP 0.81 NA 0.79 mblup NA metablup NA GBLUP+BL GBLUP+mBLUP GBLUP+metaBLUP

57 25 granjas con robot de ordeño Medida de emisión CH4 en todos los animales

58 Material and Methods Metagenome bioinformatics

59

60