class: center, middle, inverse, title-slide # Data from dirt - using Omics to understand microbial drivers in soil ### Ruth Schmidt, PhD ### Microbial Ecologist + Data Scientist ### 2020/11/13 --- <style type="text/css"> .tiny .remark-code { /*Change made here*/ font-size: 50% !important; } .small .remark-code { /*Change made here*/ font-size: 80% !important; } th { font-size: 12px; } td { font-size: 12px; } # b, strong { # color: black; # } .remark-slide tr:nth-child(2n) { background-color: white !important; } </style> ### The holobiont & hologenome theory of evolution <br> <br> > “Symbiosis is the driver of evolution.” - [Lynn Margulis](https://vimeo.com/ondemand/symbioticearthhv) <br> <p align="center"> <img width="300" src="images/micropia.png"> </p> <font size="2">Micropia Amsterdam</font> --- ### Adaptation of plant holobiont to environmental stressors - Host and its microbiota as one evolutionary unit - Microbial-driven means by which holobionts evolve within single or few generations: 1) recruitment of new microbial partners from external sources 2) amplification/reduction of the microbial partners already in place 3) horizontal gene transfer to the microbial partners already in place <br> <br> <br> <br> <br> <br> <br> <font size="2"> Margulis, L. (1991), Symbiosis as a Source of Evolutionary Innovation. MIT Press </font><br> <font size="2"> Zilber-Rosenberg I, Rosenberg E. (2008). Role of microorganisms in the evolution of animals and plants: the hologenome theory of evolution. FEMS Microbiol Rev </font> --- ### 50% human project - get in touch with your microbiome <br> <p align="center"> <img width="500" src="images/human.png"> </p> <br> <font size="2">Bäumel, S et. al. (2018), Fifty Percent Human - how art brings us in touch with our microbial cohabitants. Microb Biotechnol.</font> --- ### Volatile organic compounds (VOCs) - indicators for stress & communication signals - Key metabolites in below-ground plant-microbe interactions ([<strong style="color: #6BAED6;">Schmidt et al., 2019</strong>](https://www.nature.com/articles/s41396-019-0469-x?proof=trueSouthampton), [<strong style="color: #6BAED6;">De la Porte et al., 2020</strong>](https://linkinghub.elsevier.com/retrieve/pii/S0966842X20300524)) & in driving climate change stress (e.g. drought) responses in the plant holobiont ([<strong style="color: #6BAED6;">Schmidt et al., 2020</strong>](https://www.nature.com/articles/ismej201542)) <p align="center"> <img width="700" src="images/plant_holobiont.png"> </p> --- class: inverse, middle, center ### How would one study the role of microbial shifts, changes in genes and metabolites in the holobiont? --- ### Hypothesis & Research questions <strong style="color: #6BAED6;">Hypothesis</strong>: Drought as a stress condition - expect shifts in the microbial community structure, composition, diversity and functions that enhance plant holobiont adaptation <strong style="color: #6BAED6;">Research Questions</strong>: - How does drought influence microbial responses? - Which taxonomic groups respond to long-term and short-term water stress? - Which genes are involved in the adaptation to the stress? <strong style="color: #6BAED6;">General approach</strong>: Omics (specifically quantitative metagenomics and metabolomics) and diverse set of statistical tools* <font size="2">*ensure reproducibility (Markdown/R/data storage)</font> --- ### Experimental setup <strong style="color: #6BAED6;">Design & setup</strong>: - Randomized block design (5 replicates) of pots with two different soils (NI = non-irrigated, IR = irrigated) = 20 pots in total - After 1 month of growth, soil water holding capacity (SWHC) adjusted to 5% and 50% (Control) - Sampling of rhizosphere soil after 1 month of growth for quantitative metagenomics using internal standard method <p align="center"> <img width="600" src="images/setup.png"> </p> --- ## Metagenomic pipeline <p align="center"> <img width="650" src="images/pipeline.png"> </p> <font size="2"> Tremblay, J. et al. (2017), Chemical dispersants enhance the activity of oil- and gas condensate-degrading marine bacteria. ISME Journal </font> --- ### 1. Step: Combining raw data + metadata - Combine annotation file (TAX table and functional genes) with gene abundance file (OTU table) into one matrix - Convert OTU, TAX, sample data into one global phyloseq object using `phyloseq` package - Check object .tiny[ ```r glimpse(ps_rel@sam_data) ``` ] .tiny[ ``` Rows: 20 Columns: 3 Description.soil_type: chr [20] Description.block: chr [20] $ perc_SWHC <chr> "5", "5", "5", "5", "5", "50", "50", "50", "50", "50", "5", "5", "5", "5", "5", "50", "50", "50",… $ soil_type <chr> "NI", "NI", "NI", "NI", "NI", "NI", "NI", "NI", "NI", "NI", "IR", "IR", "IR", "IR", "IR", "IR", "… $ block <chr> "1", "2", "3", "4", "5", "1", "2", "3", "4", "5", "1", "2", "3", "4", "5", "1", "2", "3", "4", "5" ``` ] .tiny[ ```r glimpse(ps_rel@otu_table) ``` ] .tiny[ ``` Formal class 'otu_table' [package "phyloseq"] with 2 slots ..@ .Data : int [1:13108187, 1:20] 10 6 2 4 24 32 8 10 4 2 ... .. ..- attr(*, "dimnames")=List of 2 ..@ taxa_are_rows: logi TRUE ``` ] .tiny[ ```r glimpse(ps_rel@tax_table) ``` ] .tiny[ ``` Formal class 'taxonomyTable' [package "phyloseq"] with 1 slot ..@ .Data: chr [1:13108187, 1:12] "hypothetical" "COG4221" "Usp" "Sbm" ... .. ..- attr(*, "dimnames")=List of 2 ``` ] --- ### Alpha diversity (diversity within samples) significantly affected by soil type Two-way repeated measures ANOVA of alpha diversity examined by Shannon index & Simpson index - Data tested for outliers, Homogeneity of variance (Levene's Test (p > 0.05)) and normal distribution (Shapiro-Wilk’s test (p > 0.05)) using `rstatix` package .small[ ```r run_anova <- function(data, formula) { aov(formula, data) } run_anova(value ~ perc_SWHC*soil_type + Error(block)) ``` ] | | Shannon | | Simpson | | |--------------------- |---------- |--------- |---------- |--------- | | | *F* | *P* | *F* | *P* | | perc_SWHC | 0.105 | 0.75131 | 0.352 | 0.56407 | | soil_type | 12.094 | **0.00457** | 18.155 | **0.00111** | | perc_SWHC:soil_type | 0.006 | 0.9387 | 0.031 | 0.86348 | --- ### Increased diversity in soil with long-term irrigation history <p align="center"> <img width="850" src="images/alphadiv_rel_2.png"> </p> --- ### 2. step: Cleaning up data Data need to be filtered to avoid small count values, zero inflation and overdispersion. - Low count filter (removes small values caused by sequencing errors) .small[ ```r filter_taxa(ps_rel, function(x) sum(x > 4) > (0.2*length(x)), TRUE) ``` ] - Filter to remove with low Coefficient of Variation (ratio of the standard deviation to the mean) .small[ ```r filter_taxa(ps_rel, function(x) sd(x)/mean(x) > 2.0, TRUE) ``` ] <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <font size="2"> Calle ML. (2019) Statistical Analysis of Metagenomics Data. Genomics Inform.</font> --- ### 3. Unique approach: Normalization to calculate absolute abundance (counteract problem with relative data) Data need to be transformed to absolute counts by mapping number of reads per sample against genome of internal standard *Thermus thermophilus* `$$Pg = (Ps x Sa)/Sr$$` `$$Sr = Ss / Sp$$` Pg = no. of protein coding genes in sample Ps = no. of protein encoding sequences in library Sa = no. of *T. thermophilus* molecules added to sample Sr = no. of *T. thermophilus* molecules recovered Ss = no. of *T. thermophilus* reads in library Sp = no. of *T. thermophilus* protein coding genes <font size="2"> Satinsky et al. (2013) Use of internal standards for quantitative metatranscriptome and metagenome analysis. Methods Enzymol. </font> --- ### 4. Cleaned up and normalized data .small[ ```r glimpse(ps_abs@otu_table) ``` ] .small[ ``` Formal class 'otu_table' [package "phyloseq"] with 2 slots ..@ .Data : num [1:1092, 1:20] 1563 815 340 1766 1427 ... .. ..- attr(*, "dimnames")=List of 2 ..@ taxa_are_rows: logi TRUE ``` ] .small[ ```r head(ps_abs@otu_table, 5) ``` ] .small[ ``` OTU Table: [2 taxa and 20 samples] taxa are rows NI_5_1 NI_5_2 NI_5_3 NI_5_4 NI_5_5 NI_50_1 NI_50_2 gene_id_10107668 1562.643 383.2197 417.5655 607.7709 144.4909 833.5755 757.5317 gene_id_10135651 815.292 191.6099 417.5655 607.7709 216.7363 0.0000 137.7330 IR_5_2 IR_5_3 IR_5_4 IR_5_5 IR_50_1 IR_50_2 IR_50_3 gene_id_10107668 180.0927 0.0000 0 120.9514 0 0 0 gene_id_10135651 0.0000 166.8999 0 0.0000 0 0 0 ``` ] --- ### Streptomyces highly abundant in response to long-term and short-term water stress <p align="center"> <img width="800" src="images/abs_abundance_genus_bubble.png"> </p> --- ### Significant (strong) effect of soil type, as well as of SWHC and interaction of both PERMANOVA of microbial community structure | | *R<sup>2</sup>* | *P* | |--------------------- |--------- |-------- | | soil_type | 0.24576 | **0.001** | | perc_SWHC | 0.08887 | **0.024** | | soil_type x perc_SWHC | 0.07086 | **0.034** | .small[ ```r library(vegan) adonis(formula = distance(ps_abs, method = "bray") ~ soil_type * perc_SWHC, data = abs.df) Permutations = 999 ``` ] --- ### Beta diversity (diversity between samples) strongly affected by soil type Principal Coordinates Analysis (PCoA) of microbial community composition based on Bray-Curtis dissimilarity <p align="center"> <img width="550" src="images/pcoa_abs.png"> </p> --- class: inverse, middle, center # Now let's dive into the functional responses (exciting!): Genes involved in microbial response to short-term and long-term water stress --- ### `DESeq2` to analyze differentially abundant genes Advantage: integrated with `phyloseq`, better normalization .tiny[ ```r library(DESeq2) # set conditions and (multi)-factorial design ps_abs_ds = phyloseq_to_deseq2(ps_abs, ~ soil_type + perc_SWHC) # run DESeq with formula defined above and choose dispersion estimate for variance stabilizing transformation. DESeq2 estimates the width of the prior distribution from the data and therefore automatically controls the amount of shrinkage based on the observed properties of the data. ps_abs_ds = DESeq(ps_abs_ds) # fitType='local' ``` ] .tiny[ ```r library(DT) top50de <- fread(file="/Users/ruthschmidt/Dropbox/Work/INRS/Data/metagenomics_wheat_soil/Output/DE/subset_top50.txt", header = T) DT::datatable( top50de, rownames = FALSE, extensions = 'FixedColumns', fillContainer = FALSE, options = list(pageLength = 2)) ```
] --- ### Making sense of functional gene categories Heatmap of top 50 genes based on Euclidian distance-clustering using `heatmaply`
--- ### Metabolic pathway analysis using iPath and COG categories of top 50 DAG's <p align="center"> <img width="800" src="images/ipath.png"> </p> --- ### Take home messages - Soil history has a strong influence on microbial diversity, community composition and structure - Streptomyces (Actinobacteria) most abundant phyla and genera in response to water stress (resilience and VOC producers) - Functional responses: Phage related proteins, transposase, type IV secretion system (transport DNA across membrane), signal transduction system - all genes that encode for transfer of DNA/memory - Pathways: mostly primary metabolism (carbohydrate, amino acid) - common stress responses - This approach can be easily translated to other systems like the human holobiont --- class: inverse, middle, center # Metabolomics as indicator of health and disease in soil --- ### Workflow VOC data analysis .pull-left[ 1. PDMS tubes to collect VOCs (co-developed method during PhD in NL) 2. Raw data (GC-MS/QTOF, mzXML) 3. MZmine2 (open-source) for peak detection, alignments and identification of m/z values using NIST 4. Clean up data (annotations can be very messy!) 4. Statistical analysis using `MetaboAnalystR` (including time-series/two-factor) 5. Fold change, PLS-DA and VIP score, Hierarchical Clustering Heatmaps, Random forest ] .pull-right[ <img width="400" src="images/pdms.png"> ] <font size="2"> Kallenbach et al. (2015), Application of Silicone Tubing for Robust, Simple, High-throughput, and Time-resolved Analysis of Plant Volatiles in Field Experiments. The Plant Journal </font> --- class: inverse, middle, center ### THANK YOU! <img width="400" src="images/questions.png"> ### Questions?