HUB

Metagenomics 1

Amine Ghozlane

Institut Pasteur

​

Bacterial identification

1900s

Microbes cause diseases

Louis Pasteur, Robert Koch

1700s

Microscope

Anton van Leeuwenhoek

Microbes categorization

• Physical form
• In/out products

GOLDEN ERA

OF BACTERIOLOGY

1800s

Smallpox “vaccination”

(spread to EU)

1880: Optical microscopy

and Gram stain classification

1931: Electronical microscopy

Small viral particle detection

Prokaryota and Eukaryota classification

Microbes were identified by culturing and phenotyping techniques

1859: Spontaneous generation

Bacterial identification

​

1900s

Microbes cause diseases

Louis Pasteur, Robert Koch

1700s

Microscope

Anton van Leeuwenhoek

Microbes categorization

• Physical form
• In/out products

GOLDEN ERA

OF BACTERIOLOGY

1800s

Smallpox “vaccination”

(spread to EU)

1880: Optical microscopy

and Gram stain classification

1931: Electronical microscopy

Small viral particle detection

Prokaryota and Eukaryota classification

2001: ABI PRISM 3700, capable of sequencing 1M nucleotide per day.

​

1990s ---

DNA sequencing technology

Bacterial identification / characterization

Microscope

Culturing/ Phenotyping

Sequencing

Mass spectrometry

Metagenomics

To study the diversity of taxa

To study the diversity of taxa and of their associated genes/functions

To study the active fraction of the diversity

Who is there ?

• Taxonomic annotation
• Co-Abundance Gene groups (CAG)
• Metagenomic Assembled Genomes (MAGs)

​

What are they able to do ?

• Gene/protein prediction
• Functional annotation
• Metabolic network reconstruction

​

What are they doing ?

• RNA/Protein quantification

​

What is the difference between these environments ?

• Comparative metagenomics
• Quantitative metagenomics

​

Metagenomics

Metagenomics

Sample collection

DNA extraction

Functional Metagenomics

Screens capacity of an ecosystem

Targeted Metagenomics

Signature of an ecosystem

Quantitative Metagenomics

Genetic potential

Comparative Metagenomics

Comparison of ecosystems

In silico analysis

T

Targeted metagenomics

http://envgen.github.io/metagenomics.html

ITS : located between 18S and 5.8S rRNA genes

Image : Alberts Molecular Biology of the Cell 5th

50S

30S

Targeted metagenomics

Focus on one gene of the ribosomal RNA

Archaea

Bacteria

Fungi

Protozoa

[Woese et al, Nature, 1975]

1975: 16S rRNA oligomer alignment to identify 9 variable regions

1977: Using 16S/18S rRNA oligomer alignment to define the eubacteria (Bacteria), archaebacteria (Archaea) and eukaryotes (Eukaryota)

[Woese and Fox, PNAS, 1977]

A taxonomical classification based on ONE molecule !

2D electrophoretic separation to produce an oligonucleotide fingerprint

Occurence of oligomers detected at a given position in 28 organisms !

Isolate

• After Woese, classification mostly based on molecular comparison, starting with 16S

Hierarchical classification of nature initiated by Carl Linnaeus

​

• Weakly affected by horizontal gene transfer*

​

• 9 variable regions surrounded by conserved regions

​

• Universal primers**, 25 PCR cycle***

​

• Most well represented gene in Genbank

​

• Sequencing kits : V1-V3, V3-V4, V3-V5, V5-V6...

​

*Daubin et al. 2003 (Science) **Weisburg et al. 1991 (J Bacteriol.) *** Illumina protocol

Yarza et al. 2014 (Nature reviews Microbiology)

Variable regions

16S rRNA

​

​

• Weakly affected by horizontal gene transfer*

​

• 9 variable regions surrounded by conserved regions

​

• Universal primers**, 25 PCR cycle***

*Daubin et al. 2003 (Science) **Weisburg et al. 1991 (J Bacteriol.) *** Illumina protocol

16S rRNA

​

• Most well represented gene in Genbank

​

• Sequencing kits : V1-V3, V3-V4, V3-V5, V5-V6...

​

*Mysara 2017

16S rRNA

​

​

• Weakly affected by horizontal gene transfer*

​

• 9 variable regions surrounded by conserved regions

​

• Universal primers**, 25 PCR cycle***

​

• Most well represented gene in Genbank

​

• Sequencing kits : V1-V3, V3-V4, V3-V5, V5-V6...

​

16S rRNA

​

​

Streptococcus thermophilus

Streptococcus salivarius

High similarity at the genus level

16S rRNA

​

​

RAST alignment of the two genomes

S. thermophilus

ATCC 19258

~2000 proteins aligned to

S. salivarius NCTC 8618

Taxonomical annotation

SILVA [Pruesse, et al. 2007]:

• 597,607 sequences (04/2016)
• Small (16S/18S, SSU) and large subunit (23S/28S, LSU)
• Bacteria, Archaea and Eukarya
• Non redundant (Uclust 99% id)
• Based on EMBL-bank
• Updated every year more or less

​

​

Greengenes [DeSantis et al. 2006]:

• 1,262,986 sequences (last update 05/2013)
• Small subunit (16S, SSU)
• Bacteria and Archaea
• Non redundant (Uclust 99% id)
• Based on Genbank

​

​

Ribosomal Database Project [Maidak et al. 1994]:

• 3,224,600 + 108,901 sequences (last update 05/2015)
• Small subunit (16S/18S, SSU) and Fungal 28S
• Bacteria, Archaea and Fungi

​

​

Kunin et al. 2010

• 454 sequencing of V1-V2 & V8 E. coli MG1655
• Theoretical number
• 5 phylotypes for V1-V2 at 100% id
• 1 phylotype for V8
• Results
• 0.1-0.2% error probability
• 97% similarity threshold

Q20

Q20

“Group of DNA sequences that share a defined level of similarity”*

*Vetrovsky and Baldrian 2013 (Plos One)

Q30

Q30

Operational Taxonomic Unit (OTU)

• CLOSED REFERENCE CLUSTERING
• Clustering in a OTU the sequence that are similar to a reference
• Classification

​

• DE NOVO CLUSTERING
• Distance between the sequence is used to cluster sequence into OTUs

​

• OPEN REFERENCE CLUSTERING
• Closed-reference clustering followed by de novo clustering for sequence that are not similar to the reference

​

*Westcott, Schloss, 2016 PeerJ; Rideout 2014; Schmidt et al. 2015

Targeted metagenomics strategies

Hierarchical clustering

Algorithm:

• Initial n groups
• Each step:
• Merge of two group considering linkage

Distances:

• Single linkage
• the distance between two clusters is defined as the shortest distance between two points in each cluster
• Complete linkage
• the distance between two clusters is defined as the longest distance between two points in each cluster
• Average linkage
• the distance between two clusters is defined as the average distance to every point in the other cluster
• ...

​

​

​

http://www.saedsayad.com/clustering_hierarchical.htm ; https://www-users.cs.umn.edu/~kumar001/dmbook/ch8.pdf

Let’s play with

Clustering methods

Clustering criteria: 800 km

Single linkage

Hierarchical-clustering

1. Take all distance between every cities

Distance

​

Clustering criteria: 800 km

Distance

​

Clustering criteria: 800 km

Single linkage

Hierarchical-clustering

Distance

​

Clustering criteria: 800 km

Single linkage

Hierarchical-clustering

Distance

​

Clustering criteria: 800 km

Single linkage

Hierarchical-clustering

Single linkage

Hierarchical-clustering

Single linkage

Hierarchical-clustering

Compute every distance

• is really slow if you have millions of cities !!!
• requires a lot of memory

​

We need a better algorithm to do it.

SLOW !

Hypothesis

If most bases are good, most unique sequences are bad,

because good reads are all alike, but every bad read is bad in its own way. (Lost origin)

Greedy clustering

Algorithm:

• Initial n groups ordered in particular way
• Each step:
• Pick a group and compare to the reference
• If close to the reference:
• Add in reference cluster
• Otherwise:
• Add it as a reference

​

• Ordering:
• Length-based Greedy Clustering (CD-HIT, Uclust)
• Abundance-based Greedy Clustering (AGC) : “Most-Abundant-centroid”

​

Abundance-based Greedy Clustering methods

Main ideas:

​

• Abundance: Number of identical sequences in the data

​

• Sequences that have a high abundance are sequences with less errors

Abundance-based Greedy Clustering methods

New York - Los Angeles = 2451 km > 800 km

​

​

Clustering criteria: 800 km

Abundance-based Greedy Clustering methods

Abundance-based Greedy Clustering methods

New york - Los Angeles = 2451 km > 800 km

​

New york - Chicago = 713 km

​

​

​

​

Clustering criteria: 800 km

Abundance-based Greedy Clustering methods

Abundance-based Greedy Clustering methods

Abundance-based Greedy Clustering methods

Outcome:

• AGC depend on the sorting strategy (length, abundance...)

​

• The distance to the reference is guarantee…

​

• ...not the distance between sequences in the OTU

​

• AGC cost is in the worst case O(n²)…

Example :

n=200 sequences

Cost = 40000 operations <<< 8e+06 operations in hierarchical clustering

​

​

​

​

Targeted metagenomics pipeline

OTU

FastqQC

Alientrimmer,, Fastx-Toolkit, Cudadapt, …

MegaBLAST, BLAT, Vsearch, …

Flash, Pear, Pandaseq

Merging

Cleaning/Trimming

Chimera filtering

Annotation

Clustering

Mapping

Count matrix

Paired-end

Single-end

Annotation table

Dereplication

Biom

vsearch, usearch, Chimera slayer, gold database...

vsearch, usearch, cd-hit, uclust...

vsearch, usearch, cd-hit, uclust...

Singleton removal

vsearch, cd-hit, uclust...

1

2

3

4

5

6

7

Targeted metagenomics pipeline

GATTACA … AGGCTTTA

30, 31 … 20, 16, 15, 14

READ

http://drive5.com/uparse/, https://github.com/torognes/vsearch

​

Quality

Sequence

GATTACA … AGGCT

30, 31 … 20

READ

trimmed

Quality

Sequence

1) Read Trimming

Targeted metagenomics pipeline

GATTACA … AGGCTTTA

30, 31 … 20, 16, 15, 14

READ

http://drive5.com/uparse/, https://github.com/torognes/vsearch

​

Quality

Sequence

GATTACA … AGGCT

30, 31 … 20

READ

trimmed

Quality

Sequence

1) Read Trimming

2) Read Merging

GATTACA … AGGCT

Sequence

AGGCT …

Targeted metagenomics pipeline

GATTACA … GCT

GATTACA … GCT

Full length

GATTACA … GCT

GATTACA …

Prefix

GATTACA … GCT

...

Substring

3) Dereplication and singleton removal

Biological sequence X

Biological sequence Y

Chimera formed from X and Y

4) Chimera filtering

Targeted metagenomics pipeline

4) Chimera filtering

Biological sequence X

Haas et al. 2011 :

Aborted extension

Biological sequence Y

Mis-priming

Biological sequence Y

Extension

Chimera X-Y

Amplification

Targeted metagenomics pipeline

4) Chimera filtering

Biological sequence X

Haas et al. 2011 :

Aborted extension

Biological sequence Y

Mis-priming

Biological sequence Y

Extension

Chimera X-Y

Amplification

How to detect and kill the chimera ?

Targeted metagenomics pipeline

Chimera Detection

UCHIME [Edgar et al. 2011]

Previously seen sequence

Sequence S

CHUNK

Split S into chunks

CHUNK

CHUNK

CHUNK

Align and find closest pairs ( A, B)

Sequence S

B

A

Align

Compute sequence identity: A-S-B

Targeted metagenomics pipeline

OTUs

Model

Sequence

Sequence S

A. Model - Sequence <3%, Assign to OTU

B. Model - Sequence ≥3%, new OTU

OTUs

No match

Sequence

Sequence S

Add to database

5) Clustering

Sequence S

Targeted metagenomics pipeline

With the courtesy of Metagenopolis

6) Mapping

OTU

1000

The hierarchical classification of nature initiated by Carl Linnaeus today consists of eight major “ranks”

Taxonomy classification of organisms

MOST

SPECIFIC

MOST

INCLUSIVE

>94.5%

Strain

86.5%

82%

78.5%

75%

Yarza et al. 2014

Phylogeny: a complete history of life ?

Cozannet 2021

Comparing microbial communities ?

X

Y

Statistical analysis

​

Test if is the abundance of is different between X and Y

Metrics

Comparing microbial communities ?

• Evenness: Equality of the abundance

• Richness: Number of a given element

Diversity

• N: Total count of individual
• Rarefaction: a type of normalisation : Rarefy to the same number of individual

​

N = 33

X

​

N = 16

Y

Diversity

• N: Total count of individual
• Rarefaction: a type of normalisation : Rarefy to the same number of individual
• S = number of species = richness = number of object > 0

S = 4

N = 16

S = 3

N = 16

X

Y

RAREFACTION = “DOWNSIZING”

Alpha diversity

• S = number of species = richness = number of object > 0
• Alpha diversity:
• Condition 1 : α₁ = mean(S_A, S_B) = 3.5
• Condition 2 : α₂ = mean(S_C, S_D) = 5

​

A

B

C

S = 3

N= 10

S = 4

N = 10

S = 2

N = 10

CONDITION 1

CONDITION 2

D

S = 8

N = 10

Gamma diversity

• S = number of species = richness = number of object > 0
• Gamma diversity:
• Condition 1 : ɣ₁ = S₁ = 5
• Condition 2 : ɣ₂ = S₂ = 8

A

B

C

S = 3

N= 10

S = 4

N = 10

S = 2

N = 10

CONDITION 1

CONDITION 2

D

S = 8

N = 10

Beta diversity

• S = number of species = richness = number of object > 0
• Beta diversity:
• Condition 1 : β₁ = - 1 = 0.43

​

• Condition 2 : β₂ = 0.6

A

B

C

S = 3

N= 10

S = 4

N = 10

S = 2

N = 10

CONDITION 1

CONDITION 2

D

S = 8

N = 10

ɣ₁

α₁

Shannon

• Indice de Shannon:

​

-(4/16*log(4/16)+5/16*log(5/16)+2/16*log(2/16)+5/16*log(5/16))

S = 4

N = 16

S = 3

N = 16

X

Y

Shannon = 1.33

Shannon = 1.08

Bray-Curtis dissimilarity

A

B

S_A = 3

N= 10

S_B = 4

N = 10

• C_AB: The sum of the lesser values for the species found in each site.

​

• C_AB = 3 + 0 + 1 + 0 + 0 = 4

​

• BC_AB = 1 - (2*C_AB) /(N_A + N_B) = 1 - ( 2 * 4 ) / ( 20 ) = 0.6

​

Bray-Curtis dissimilarity

A

B

S_A = 3

N= 10

S_B = 4

N = 10

• C_AB: The sum of the lesser values for the species found in each site.

​

• 0 = No dissimilarity = same number of each type of species

​

• 1 = Complete dissimilarity = share non of the same type of species

​

• The Bray-Curtis dissimilarity assumes that the sites are of equal size.

Otherwise a larger size (N) in one site would be misleading because the difference is not their composition but their size.

Unifrac distance

Multiple sequence

Alignment

Unifrac

Muscle, Mafft

Tree construction

Rooting tree

Fasttree, raxml…

• Lozupone and Knight 2005

​

• Introduce the sequence similarity parameter using a phylogenetic distance

​

​

Principal Coordinates Analysis (PCoA)

• PCoA attempts to represent inter-object (dis)similarity in a low-dimensional.

​

• Rather than using raw data, PCoA takes a (dis)similarity matrix as input.

​

• It is conceptually similar to principal components analysis (PCA) which preserve Euclidean and χ2 (chi-squared) distances between objects, respectively; however, PCoA can preserve distances generated from any (dis)similarity measure allowing more flexible handling of complex ecological data.

​

Rarefaction curves

Krona plot

16S – Challenges

*Vetrovsky and Baldrian 2013 (Plos One), Klappenbach et al. 2001 (Nuc Acid Res)

**Acinas et al. 2004 (J Bacteriol)

***Stoddard et al. 2015 (Nucl Acid Res)

• Copy number varies in the genomes (from 1 to 15)*
• 16S sequence variants in the same specie and even genome** -> impact diversity

16S – Challenges

*Vetrovsky and Baldrian 2013 (Plos One), Klappenbach et al. 2001 (Nuc Acid Res)

**Acinas et al. 2004 (J Bacteriol)

***Stoddard et al. 2015 (Nucl Acid Res)

• Copy number varies in the genomes (from 1 to 15)*
• 16S sequence variants in the same specie and even genome** -> impact diversity

NO !

16S – Challenges

*Vetrovsky and Baldrian 2013 (Plos One), Klappenbach et al. 2001 (Nuc Acid Res)

**Acinas et al. 2004 (J Bacteriol)

***Stoddard et al. 2015 (Nucl Acid Res)

• Copy number varies in the genomes (from 1 to 15)*
• 16S sequence variants in the same specie and even genome** -> impact diversity

OK ! but in absolute count

​

NO !

Solution?

• rrnDB: ribosomal RNA operon copy number database***

​

BIOM format

Motivation:

• Encapsulation of the whole project (count table, annotation, metadata...)
• Efficient storage (version 2.0: HD5 binary format)
• Compatibility between software

​

​

Annotation

Metadata

Count

shaman.pasteur.fr

Better statistics

Simplicity

Reproducible research

Raw data submission

71 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Unique key associated to your email

Parameters are saved with your key

We do not store your reads after calculations unless:

• there is crash and you might come discuss with us: shaman@pasteur.fr

-> crashed dataset are deleted regularly

Except ITS databases: Unite, Findley, Underhill

Raw data submission

72 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Global report

Detailed process

Download available

A mail is also automatically sent

Processed data submission

73 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Tables

BIOM / Epi2me (nanopore)

Key

Exemple datasets

Processed data submission

74 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

% of OTU annotated

Download available

Unifrac enabled

​

Processed data submission

75 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

DESeq2 Metagenomics

16S : McMurdie, Holmes, Plos Comput Biol,2014

WGS : Jonsson, BMC Genomics, 2016

​

76 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Check our publication in BMC Bioinfo !

Number of observations

Sparsity ratio

Modified normalizations

78 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Data filtering

79 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Statistical modelling

Automatically determined from the statistical model

31 interactives visualizations

81 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Diagnostic plots

How good is my normalisation ? modelisation ? effect size ?

82 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Significant features

83 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Global visualisations

84 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Global visualisations

abundance tree

85 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

Global visualisations

86 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

An. stephensi Prevalence GS

Cedecea R² =15%

Serratia R² =26%

Contrast comparison

87 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 25/06/2021

shaman.pasteur.fr

« There is no disputing the importance of statistical analysis in biological research, but too often it is considered only after an experiment is completed, when it may be too late. »

​

Raw data submission

(targeted metagenomics)

Targeted metagenomics at Pasteur

shaman.pasteur.fr

Annotation

Count matrix

shaman.pasteur.fr

target

M2IPFB

OTU

Q20

Id 0.75 on both strand,

take best hit only,

Database is in database folder

Merging

Cleaning/Trimming

Chimera filtering

Annotation

Clustering

Mapping

Count matrix

Paired-end

Annotation table

Dereplication

De novo chimera filtering

Clustering based on abundance at Id 0.97 both strand

Dereplication fulllength on both strand

Singleton removal

Min abundance > 10

min overlap 40,

max diffs 15

Group all sequences in one file

Each amplicon origin’s is specified !

Global alignment at Id 0.97 both strand

SHAMAN

M2BI

OTU

Q20

Id 0.75 on both strand,

take best hit only,

Database is in database folder

Merging

Cleaning/Trimming

Chimera filtering

Annotation

Clustering

Mapping

Count matrix

Paired-end

Annotation table

Dereplication

De novo chimera filtering

Clustering based on abundance at Id 0.97 both strand

Dereplication fulllength on both strand

Singleton removal

Min abundance > 10

min overlap 40,

max diffs 15

Group all sequences in one file

Each amplicon origin’s is specified !

Global alignment at Id 0.97 both strand

For the practical sessions

you will need to use a VPN !

Option 1: Install a remote access software

How (if not installed) ?

Go to the following address : http://connect.pasteur.fr/

and then get logged with your usual Pasteur ID and password

​

Scroll down, until you find :

​

​

Choose the suited remote access client (depending on your OS)

and install it !

Once installed, you should be able to open the software and log-in using your usual Pasteur IDs

For the practical sessions

you will need to use a VPN !

Option 2: directly use the connect.pasteur.fr interface

Go to the following address : http://connect.pasteur.fr/

and then get logged with your usual Pasteur ID and password

​

​

​

​

​

Enter the VM address :

desktop.pasteur.fr

Practical session - Command line on virtual machine

Ubuntu virtual machine accessible with web browser or using VMware

https://desktop.pasteur.fr

​

​

​

Practical session - Command line on virtual machine

Ubuntu virtual machine accessible with web browser or using VMware

https://desktop.pasteur.fr

​

​

​

Practical session

• Right-click on the desktop -> click on “open in terminal”

​

​

​

​

M2BI

• You must develop a program that clusterize 16S sequence into OTU

• Zymobiomics mixture: https://www.ozyme.fr/gammes/zym/pdf/zym_zymobiomics_brochure.pdf

WARNINGS

​

• Clustering is performed on the sequence in one step
• You must identify from which sample come your amplicon

​

@M01626:316:000000000-AY73Y:1:1101:15636:1045 1:N:0:38

>M01626:316:000000000-AY73Y:1:1101:11704:11521:N:0:38;sample=1ng-25cycles-1;

• Count table

• Annotation table

Amplicon Fasta:

A bacteriocin from epidemic Listeria strains alters the host intestinal microbiota

• Javier Pizarro-Cerda – Juan Jose Quereda Torres

101 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 28/06/2016

A bacteriocin from epidemic Listeria strains alters the host intestinal microbiota

• Javier Pizarro-Cerda – Juan Jose Quereda Torres

102 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 28/06/2016

A bacteriocin from epidemic Listeria strains alters the host intestinal microbiota

WT

Delta

Delta-compl

-48h

-24h

6h

24h

T0

T2

T3

16S : V1-V3

​

• Impact on major communities (Which ? When ? How much ?)
• Quereda, Dussurget, Nahori, Ghozlane, Volant, Dillies, et al.,

PNAS 2016

​

​

​

​

103 ∙ Amine Ghozlane • SHAMAN : Shiny Application for Metagenomic ANalysis ∙ 28/06/2016

Data quality check

Evaluation of the read quality by plotting the Phred scores

and the distribution of mean read quality

Sequencing quality tends to decrease at the end of the reads

- > Trimming step

Data quality check

Evaluation of the read quality by plotting the Phred scores

and the distribution of mean read quality

Sequencing quality tends to decrease at the end of the reads

- > Trimming step

Data quality check

These patterns are expected due to the ‘not so random’ primers that will fix on regions of the genome with biaised GC content.

Data quality check

This profil means that some reads have an intermediate length (excess of intermediate-length reads)

-> The aligner can easily deal with it

This profil means that some reads are duplicated

Data quality check

What about metagenomics real life data now ?

Data quality check

Data quality check

Data quality check

Data quality check

​

Do you think it is worth trimming

the bad quality reads ?

​

You can run your own fastqc analysis in the terminal

Targeted metagenomics pipeline

Where to find the data ?

​

​

ctp_1.tar.gz

​

​

c3bi.pasteur.fr/m2p7

and

Click on cours1

Metagenomics

application to

Image : http://phdcomics.com/comics/archive.php?comicid=1874

Mapping

*Edgar et al. 2013 (Nature methods)

Isolated singleton

“Bad” singleton

Recaptured singletons

Kunin et al. 2010

• 454 sequencing of V1-V2 & V8 E. coli MG1655
• Theoretical number
• 5 phylotypes for V1-V2 at 100% id
• 1 phylotype for V8
• Results
• 0.1-0.2% error probability
• 97% similarity threshold

“Group of DNA sequences that share a defined level of similarity”*

*Vetrovsky and Baldrian 2013 (Plos One)

Operational Taxonomic Unit (OTU)

Loosely based on Stackebrandt and Ebers 2006

“Group of DNA sequences that share a defined level of similarity”*

*Vetrovsky and Baldrian 2013 (Plos One)

Operational Taxonomic Unit (OTU)

Johnson 2019