1 of 27

Ancient Metagenomics

��Emrah Kırdök, PhD

2 of 27

Overview

Ancient Metagenomics:

Metagenomics
Methods on Ancient metagenomics research

Ancient Diseases:

How metagenomics helps us to understand ancient diseases?

3 of 27

Ancient Metagenomics

4 of 27

Metagenomics

Most of the microorganisms are uncultivable.

The aim is to understand:

Who are they?
What are they doing?
How they interact with each other?

Metagenomics consists of a series of methods that helps us to understand microbial diversity in a particular niche. Traditional microbiological methods allows us to understand to a point.

However, less then 5% of the micro organisms are culturable using the traditional microbiology methods. That means we could not access a big part of the microbial diversity.

Metagenomics helps us to access the microbial species that we can not culturable using traditional microbiology methods and identify them through DNA sequence traces.

The aim of the metagenomics are could be summarised down to three points:

Who are they?

In metagenomics we would like to understand the microbial composition of the ecological niche of interest.
This step is also called as taxonomical classification

What are they doing?

Functional classification

How they interact with each other?

We could compare different ecological niches and how they interact.

Think about human microbiome. Every body part has its own niche but they are somewhat connected.

5 of 27

The bowl of candies could represent an ecological niche

6 of 27

The NMDS visualisation of Human Microbiome Project Samples. PC 1 and 2 (a), PC 1 and 3 (b)

7 of 27

Metagenomics

8 of 27

Reference genome

DNA sequences

%95

%97

%99

In a typical ancient demography study, we randomly amplify DNA sequences from a historical specimen and sequence. The result is a soup of DNA sequences called as a DNA library.

General procedure is to align those sequences to the host organism. In our case let say we are working with human reference genome. The idea is to find the best possible genomic locations for those sequences.

Generally we calculate a similarity value between sequences and possible genomic locations. Best similarity value means we have a match to for this sequence. This procedure is called as mapping or alignment

This step is important because we would like to compare chromosomal locations between different samples. So we need to be confident with our alignment. Thus we choose a place that has the best homology.

9 of 27

Reference genome 1

Reference genome 2

Reference genome 3

DNA sequences

%99

%97

%98

%99

%97

10 of 27

Metagenomics

Databases

RefSeq
HOMD

Human oral microbiome database

HMP

Human microbiome database

11 of 27

Reference genome 1

Reference genome 2

Reference genome 3

Root

Species

Genus

98%

Reference genome 4

95%

93%

Root - Domain – Phylum - …. – Family - Genus - Species

Family

We have demonstrated competitive mapping. But we should include taxonomy information, otherwise we could not place the reads on a taxonomical node.

Taxonomy is a tree like hierarchical structure. The base node is named as root. All the lineages diverge from that point.

Taxonomy information is needed to curate reference genome databases, and place DNA sequences on higher taxonomical nodes if needed.

Which taxonomic clade are the reads fall? We align our DNA sequences to reference genomes. These genomes have taxonomy information that defines their place in the taxonomy tree. If we could map a DNA sequence to only one reference genome with a high similarity, that means we find a taxonomy lineage for this sequence.

What if we map aDNA sequence to more then one reference genome? Then we need to analyse the alignments on a taxonomy tree. So, it is possible to place a DNA sequence anywhere in the taxonomical tree.

12 of 27

Species A

Species B

7

6

13 of 27

Ancient Metagenomics

The timescale of ancient DNA studies (Hofreiter et al., 2014)

14 of 27

< %1 maps to host genome

(but really depends on the conservation status)

%20 - %34 identified

%65 - %79 remains unaligned

Composition of an aDNA library

Host microbiome

Environmental microbiome

15 of 27

Deamination patterns observed in aDNA sequences

16 of 27

Ancient Metagenomics

Not all species exists in our databases

Database bias to well studies microorganisms

17 of 27

Ancient Metagenomics

How metagenomic methods helps us to understand ancient microbiome?

18 of 27

Sampling

Ancient calculus

Teeth

Masticates

Gut microbiome in mummified tissues

Long bones

(specific lesions)

19 of 27

The ancient metagenomic process starts with a usual ancient demography analysis. We extract DNA sequences from historical specimens and select the sequencing strategy.

In amplicon sequencing, we amplify marker sequences that distinguish between species. For bacteria, that marker sequence is generally 16S rDNA sequence. Using universal primers that targets this sequences, we amplify DNA sequences and then use next generation sequencing technology. After sequencing, we align marker sequences back to their references. Then we count the aligned sequences. Since we know the copy number of each gene, we could normalise the counted value and find the abundance of each microbe.

In shotgun sequencing, we randomly amplify what we have found in the historical sample. We do not seek for any particular markers. In modern metagenomic studies, generally we chemically or physically break DNA strand to form smaller fragments. However in ancient DNA studies we do not to do this since we have already fragmented DNA.Then we align our sequences to the reference genome database.

There are other strategies as well. If we could create a capture panel, we could easily enrich microbial DNA or a specific pathogenic genome.

20 of 27

Tools in metagenomics

Taxonomical profiling
Taxonomical classification

21 of 27

Metaphlan2

A set of marker database:

Clade specific marker database

15 million markers
180 markers for each species

Can not use all the sequences

22 of 27

Malt

Alignment based classification tool

23 of 27

Kraken2

Uses k-mer profiles instead of aligning
Very fast and sensitive but precision is low.

24 of 27

Taxonomical Classification

Root

Species

Genus

Family

Y. pestis

Y. pseudotuberculosis

98%

96%

99%

94%

Yersinia pestis

95%

98%

99%

98%

Yersinia

Lowest common ancestor method.

The current paradigm in LCA methodology is to label a DNA read with an appropriate taxonomical node.

Let’s think that we are doing our metagenomic search through homology. Tools often set a homology threshold to place a DNA read into a taxonomical node.

In SC1 lets think that our DNA reads matched into a specific species reference genome. Using a 97% similarity thresold, we could confidentially map the DNA read to a specific species taxonomy node. Since the read has aligned only the reference genomes belonged to this species.

However most of the time that is not the case. DNA reads often aligns more then one species reference genomes with a high homology. In SC2, we end up with two species in the same genus.

In this case, LCA of this read is the genus taxonomy node.

25 of 27

What is the result?

Species	Absoute Abundance	Relative Abundance
Species A	23	0.12
Species B	45	0.24
Species C	76	0.4
Species D	45	0.23

26 of 27

Sample

DNA

Metagenomics

Microbial Abundances

Microbial Diversity

Pathogen Genomics

27 of 27

Diversity analysis

Alpha diversity

Number of species observed in a particular niche

Beta diversity

Bray-Curtis distance