bladder campion and soil copper pollution

On the ability of Silene vulgaris to deal with copper contaminants in soil

Aim of the project

To improve de novo transcriptome assembly of S. vulgaris
To compare transcriptomes of two S. vulgaris populations
To find key genes responsible for copper tolerance

Input: RNA-seq datasets (24 samples)

S. vulgaris population Stranska skala (nonmetallicolous population from Brno, Czech Republic: N 49.190146, E 16.675319)

S. vulgaris population Lubietova (metallicolous population from Slovakia: N 48.747893, E 19.385472)

The flowchart of de novo transcriptome assembly and differential gene transcription analyses

Input: 48 fastq files (paired end RNA sequencing, Illumina HiSeq, 24 samples )

Filtering - the downstream analyses of the raw Trinity's output

1.) input: 289 610 sequences (raw output of Trinity de novo assembly)

Tool: RSEM with bowtie2 RSEM package RSEM article
Purpose: to identify all misassembled and nonsense contigs
Description: the re-mapping of all reads (the concatenated fastq files) to the de novo assembled transcriptome (reference)
Solution: all contigs with TPM value equal 0 or/and IsoPct value equal 0 are excluded
Note: TPM - transcripts per million TPM clear explanation
Note: IsoPct - isoform percentage Optimizing assembly
Output: 186 921 sequences

2.) input: 186 921 sequences (output from the first step )

Tool: CAP3 software with the defualt setting CAP3 package CAP3 article
Purpose: to eliminate the number of duplicates, isoforms and splicing forms
Description: the generation of consensus sequences based on the multiple sequence alignments of contigs
Solution: to keep only one consensus sequence instead of multiple isoforms
Note: Isoform - de Bruijn graph per expressed gene Trinity assembly method
Output: 127 373 sequences

3.) input: 127 373 sequences (output from the second step )

Tool: blastn, e-value cut-off 1e ^-10
Purpose: to remove possible contaminants - sequences from plastid genomes, plant pathogens, known transposable elements, S. vulgaris 18S rRNA, known plant repeats; and to select one representative sequence per group of very similar sequences (e-value less than ^-50 )
Description: if a contig is not unique, select the longest sequence from a group of similar sequences based on bd-blastn; if the unique or selected sequence does NOT give the best blastn hit with a repeat/pathogen/TE/chlorplastome/mitochonriome/rRNA sequence, the sequence is kept among filtered de novo assembled transcripts and assumed as a product of protein coding nuclear gene transcription
Solution:

blastn against databases of plastid genomes, plant pathogens, repeats
bi-derectional blastn (queried sequences are the same as a set of sequences used to build a reference database)

Note: S. vulgaris reference plastid genomes mitochondrial genome chloroplast genome
Note: plant repeat element database mips
Note: database of ribosomal RNA sequences SILVA db
Note: database of repeats and transposable elements repbase
Note: database of plant pathogen's sequences plant pathogens DB
Output: 53 107 sequences

BUSCO results for particular steps and attempts during the assembly

More info about BUSCO: BUSCO publication and source code

BUSCO results of our best transcriptome assembly and other publicly available assemblies

The used plant genomes downloaded from Phytozome DB

Silene vulgaris publicly available transcriptome Taylor Lab

Reload

The best Athaliana blastX hits (max=3, e-value < 0.001) with Silene vulgaris transcriptome:

contig1

Results of differential gene expression analysis

de novo assembly

Raw output from Trinity:

SV_rawOutput_Trinity.fa

Filtered assembly estimated as the best one:

SV_filtered_53107seqs_Trinity.fa

SAF files from the best filtered assembly dedicated for FeatureCounts tool:

SV_filtered_53107seqs_Trinity.saf

Protein prediction in TransDecoder tool

gff3 file of all predicted proteins (longer than 30 AAs):

Silene_vulgaris.gff3

All predicted ORFs (longer than 30AAs), fasta file:

Svulgaris_allORFs_nt.fa

All predicted proteins (longer than 30AAs), fasta file:

Svulgaris_proteins_AA.fa

Filtered set of proteins (one protein per contig), fasta file:

Svulgaris_proteins_filteredSet.fa

Domain prediction

Simplified table of found domain(s) per protein (input: the filtered set of proteins):

SV_filtered_domainPerSeq.csv

Differential gene transcription analyses:

Raw count tables per tissue (12 samples):

Svulgaris_roots_RawCounts_48103seqs.csv
Svulgaris_leaves_RawCounts_49617seqs.csv

On the ability of Silene vulgaris to deal with copper contaminants in soil

Aim of the project

To improve de novo transcriptome assembly of S. vulgaris

To compare transcriptomes of two S. vulgaris populations

To find key genes responsible for copper tolerance

Input: RNA-seq datasets (24 samples)

S. vulgaris population Stranska skala (nonmetallicolous population from Brno, Czech Republic: N 49.190146, E 16.675319)

S. vulgaris population Lubietova (metallicolous population from Slovakia: N 48.747893, E 19.385472)

The flowchart of de novo transcriptome assembly and differential gene transcription analyses

Input: 48 fastq files (paired end RNA sequencing, Illumina HiSeq, 24 samples )

Filtering - the downstream analyses of the raw Trinity's output

1.) input: 289 610 sequences (raw output of Trinity de novo assembly)

Tool: RSEM with bowtie2 RSEM package RSEM article

Purpose: to identify all misassembled and nonsense contigs

Description: the re-mapping of all reads (the concatenated fastq files) to the de novo assembled transcriptome (reference)

Solution: all contigs with TPM value equal 0 or/and IsoPct value equal 0 are excluded

Note: TPM - transcripts per million TPM clear explanation

Note: IsoPct - isoform percentage Optimizing assembly

Output: 186 921 sequences

2.) input: 186 921 sequences (output from the first step )

Tool: CAP3 software with the defualt setting CAP3 package CAP3 article

Purpose: to eliminate the number of duplicates, isoforms and splicing forms

Description: the generation of consensus sequences based on the multiple sequence alignments of contigs

Solution: to keep only one consensus sequence instead of multiple isoforms

Note: Isoform - de Bruijn graph per expressed gene Trinity assembly method

Output: 127 373 sequences

3.) input: 127 373 sequences (output from the second step )

Tool: blastn, e-value cut-off 1e -10

Purpose: to remove possible contaminants - sequences from plastid genomes, plant pathogens, known transposable elements, S. vulgaris 18S rRNA, known plant repeats; and to select one representative sequence per group of very similar sequences (e-value less than -50 )

Solution:

blastn against databases of plastid genomes, plant pathogens, repeats

bi-derectional blastn (queried sequences are the same as a set of sequences used to build a reference database)

Note: S. vulgaris reference plastid genomes mitochondrial genome chloroplast genome

Note: plant repeat element database mips

Note: database of ribosomal RNA sequences SILVA db

Note: database of repeats and transposable elements repbase

Note: database of plant pathogen's sequences plant pathogens DB

Output: 53 107 sequences

BUSCO results for particular steps and attempts during the assembly

More info about BUSCO: BUSCO publication and source code

BUSCO results of our best transcriptome assembly and other publicly available assemblies

The used plant genomes downloaded from Phytozome DB

Silene vulgaris publicly available transcriptome Taylor Lab

The best Athaliana blastX hits (max=3, e-value < 0.001) with Silene vulgaris transcriptome:

Results of differential gene expression analysis

de novo assembly

Raw output from Trinity:

Filtered assembly estimated as the best one:

SAF files from the best filtered assembly dedicated for FeatureCounts tool:

Protein prediction in TransDecoder tool

gff3 file of all predicted proteins (longer than 30 AAs):

All predicted ORFs (longer than 30AAs), fasta file:

All predicted proteins (longer than 30AAs), fasta file:

Filtered set of proteins (one protein per contig), fasta file:

Domain prediction

Simplified table of found domain(s) per protein (input: the filtered set of proteins):

Differential gene transcription analyses:

Raw count tables per tissue (12 samples):

regularized log normalized (rlog) tables per tissue (12 samples):

lower cut-off: mean coverage more than 15 mapped reads per contig, median coverage more than 9 mapped reads per contig

upper cut-off: mean less than 100 000 mapped reads per contig, median less than 100 000 mapped reads per contig

DESeq2 results

roots - SS versus LUB:

ref: SS population , query: LUB population

roots - LUB control versus LUB copper:

ref: LUB control, query: LUB copper enriched soil

roots - SS control versus SS copper:

ref: SS control, query: SS copper enriched soil

leaves - SS versus LUB:

ref: SS population, query: LUB population

Novel transposons reconstructed in Silene latifolia :

fasta file of the reconstructed TEs, for details see Kralova et al., 2014

Table with raw read counts mapped onto repeats and transposons

Discussion and highlights

Tool: blastn, e-value cut-off 1e ^-10

Purpose: to remove possible contaminants - sequences from plastid genomes, plant pathogens, known transposable elements, S. vulgaris 18S rRNA, known plant repeats; and to select one representative sequence per group of very similar sequences (e-value less than ^-50 )