In this tutorial, we’ll walk through extracting and assembling the mitochondrial genome of Dodonaea viscosa using Oxford Nanopore reads. Measures are always taken to remove organelles prior to sequencing, though there are usually enough long reads to create single-contig, complete assemblies. To avoid confounding our assemblies, we will be first employ strategies to remove NUMTs (nuclear mitochondrial DNA segments) and NUPTs (nuclear plastid DNA segments) from our reads. What are NUMTs and NUPTs?
- NUMTs (Nuclear Mitochondrial DNA) are segments of mitochondrial DNA that have been inserted into the nuclear genome and can vary greatly across species. These can be present in multiple copies with varying degrees of sequence divergence from the functional mitochondrial genome.
- NUPTs (Nuclear Plastid DNA) are the plastid equivalent.
Then I will show you how to install and assemble these reads with five different assemblers.
- Unicycler – Works best with hybrid data, but in long-read-only mode may struggle with noisy reads and NUMTs when assembling small organellar genomes.
- Miniasm – Assembles quickly but without error correction, so mitochondrial contigs may be fragmented or inaccurate without correction.
- Flye – Probably the best at recovering circular mitochondrial genomes from whole-genome reads, even in the presence of NUMTs, due to good repeat handling and built-in polishing.
- Canu – Accurate but much slower; assembles mitochondrial genomes well if coverage is high, can struggle with NUMTs.
- Raven – Fast and memory-efficient; often produces clean mitochondrial contigs even from mixed read sets, though may miss complex repeats or low coverage regions.
Setup: Organize your workspace
Where are your nuclear-targeted or other long reads?
I have a set of reads I have already been trimmed for adapters, sequence quality, and a minimum of 5kb in length. Keeping your input files at the top of your repo is good practice.
1
/work/gif3/sharu/viscosa/03_Porechop/trimmed_5kplus.fq
Where is your nuclear genome?
1
MitoRemovedpolished_3.fasta
Where on the HPC are you storing these files for future reference?
1
/work/gif3/masonbrink/USDA/04_MitochondrialIsolationAndAssembly
Find related species with an assembled mitochondrial genome
To aid in assembly and annotation, find mitochondrial genomes from related species.If not from the same genus, then the same family, etc. My reads are from Dodonaea viscosa, a plant species in the sapindaceae.
-
Search the NCBI Nucleotide database for “Sapindaceae mitochondrion genome.”
-
Filter for plants and mitochondria.
-
Download sequences for related species. Some examples in my case:
- Xanthoceras sorbifolium
- Nephelium lappaceum
- Litchi chinensis
- Sapindus mukorossi
Once you find a few assemblies of related species, you can visualize how they are related by plotting ‘genus species’ names into the NCBI Taxonomy Browser tool: Common Tree
https://www.ncbi.nlm.nih.gov/Taxonomy/CommonTree/wwwcmt.cgi
Identifying organellar genomic reads from nuclear-targeted reads
Install Minimap2
Map reads to related species’ mitochondrial genome
1
sh runMinimap.sh trimmed_5kplus.fq RelatedNCBIMitochondrialGenomeSequences.fasta
runMinimap.sh
Install Seqtk
1. Extract reads from related mitochondrial genome mapping with at least 2kb alignment length to the using Seqtk
The length of the alignment filter is to get rid of any nuclear integrated mitochondrial genomic fragments. If your genome assembly has very few or small NUMTs or NUMPs then, this is likely the easiest way to obtain an organelle-targeted set of reads from a nuclear-targeted set of reads.
1
less trimmed_5kplus.fq_RelatedNCBIMitochondrialGenomeSequences_minimap2.paf |awk -F"\t" '($4-$3)>2000 {print $1}' |sort|uniq |seqtk subseq trimmed_5kplus.fq.gz - >MitoNanopore2k.fastq
2. Or we can map the reads to the nuclear genome and extract only those that did not map.
The thought here is that if there is not too much contamination in the reads, we will be able to eliminate all reads that are nuclear and keep only low-quality reads, organelle reads, and contamination reads to use for organelle genome assembly.
Softlink genome, reads and mapping script
1
2
3
ln -s ../../03_DononeaViscosaDeposition/MitoRemovedpolished_3.fasta
ln -s ../trimmed_5kplus.fq
ln -s ../runMinimap.sh
Map reads to nuclear genome
1
echo "sh runMinimap.sh trimmed_5kplus.fq MitoRemovedpolished_3.fasta" >Map2Nucleus.sh
Get names of all reads in the dataset
1
awk 'NR % 4 == 1' trimmed_5kplus.fq |sed 's/^@//g' |awk '{print $1}' >AllTrimmedReadNamesFormatted.list
Get the names of the reads that are not mapping to the nuclear genome
1
cat AllTrimmedReadNamesFormatted.list <(awk -F"\t" '$12==60 && $10>500' {print $1}' trimmed_5kplus_MitoRemovedpolished_3_minimap2.paf|uniq ) |sort |uniq -c |awk '$1==1 {print $2}' >OrganelleReads.list
Extract those reads that did not map
1
seqtk subseq trimmed_5kplus.fq OrganelleReads.list >OrganelleReads.fastq
3. Or we can identify reads that have mitochondrial genes using miniprot and assemble those reads.
Here we will use the predicted proteins from the Sapindus mukorossi mitchondrial genome to identify mitochondrial reads among D. viscosa’s nuclear-targeted fastq file.
Note that in my case this generated a small number of reads that were not of sufficient depth to split, so I ran assemblies on this subset of reads without splitting the fastq files.
Convert fastq to fasta
1
micromamba activate seqtk; seqtk seq -a trimmed_5kplus.fq >trimmed_5kplus.fasta
Map proteins to the reads
1
echo "ml micromamba; micromamba activate miniprot;miniprot -t 64 -S trimmed_5kplus.fasta NamedSmukorossiProteins.fasta >SmukrossiProteins2OrganelleReads.paf " >miniprotToTrimmed5kplus.sh
Grab all read names that have a mitochondrial protein alignment
1
less SmukrossiProteins2OrganelleReads.paf |cut -f 6 |sort |uniq >MiniprotOrganelleReads.list
Extract the reads with seqtk
1
seqtk subseq trimmed_5kplus.fq MiniprotOrganelleReads.list >MiniprotOrganelleReads.fastq
Create subsets of reads from different positions in the fastq file using Trycycler
Install Trycycler
``` micromamba create -c bioconda -c conda-forge -n trycycler trycycler micromamba activate trycycler ```
Here trycycler will generate 12 fastq files in the read_subsetNuclearClean and read_subsets folders, that are close to equal in size. If depth is sufficient, then you will get split files without any redundant read overlap.
1
2
trycycler subsample --reads OrganelleReads.fastq --out_dir read_subsetNuclearClean -t 36
trycycler subsample --reads MitoNanopore2k.fastq --out_dir read_subsets -t 36
Alternatively generate random subsets with Seqtk by changing the seed number (-s) and proportion.
1
2
3
4
# By modifying '-s' and your output filename, you can generate different subsets of 10% of the reads.
seqtk sample -s100 MitoNanopore2k.fastq 0.1 > subset_00.fastq
seqtk sample -s101 MitoNanopore2k.fastq 0.1 > subset_01.fastq
etc...
Installing software for creating assemblies from read subsets
After choosing your method to split the fastq files, we will use these to generate multiple independent assemblies. Here I will be using Unicycler, Miniasm, Raven, CANU, and FLYE to generate different assemblies. One or a few assemblers may be fine for your purposes.
Install Unicycler assembler
Install Flye assembler
Install Miniasm assembler
Install Raven assembler
Install CANU assembler
Install Circlator
Circlator will rotate circular assemblies so that the genome starts at a standard gene (dnaA usually ) for better comparisons to published assemblies.
1
2
3
4
5
6
7
8
# Use multiple assemblers to generate genome assemblies of each read subset
To prepare this tutorial I created >100 assemblies, though you may achieve good results with just one run of it. Sometimes the filtered fastq file is too small to split, and was thus used as a whole. You also do not need to install all of these assemblers, one or a few may be fine.
**Generate Flye Assemblies**
```bash
for f in *fastq; do echo "ml micromamba; micromamba activate flye; flye --nano-raw $f --threads 36 --out-dir ${f%.*}FlyeOut --genome-size 800000 --meta --scaffold" -m 2000 ; done >flye.sh
Generate Miniasm Assemblies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
for f in *fastq; do echo "sh AssembleMitoMiniasm.sh "$f" "${f%.*}"_MiniasmOut";done >miniasmAssemblies.sh
**AssembleMitoMiniasm.sh**
<details> <summary>Details</summary>
#!/bin/bash
#Ensure script stops on errors
set -e
#Check if correct number of arguments are provided
if [[ $# -ne 2 ]]; then
echo "Usage: sh assemble_mito.sh <reads.fastq> <output_prefix>"
exit 1
fi
#Assign command-line arguments to variables
READS="$1"
PREFIX="$2"
#Create an output directory
OUTDIR="${PREFIX}_output"
mkdir -p "$OUTDIR"
#Check if input file exists
if [[ ! -f "$READS" ]]; then
echo "Error: File '$READS' not found!"
exit 1
fi
#Convert FASTQ to FASTA and save in the output directory
echo "Converting FASTQ to FASTA..."
awk 'NR%4==1 {print ">" substr($0,2)} NR%4==2 {print}' "$READS" > "$OUTDIR/${PREFIX}_reads.fasta"
#Run minimap2 for read overlap
echo "Running minimap2..."
ml micromamba;micromamba activate minimap2;minimap2 -x ava-ont -t 36 -k 19 -w5 "$OUTDIR/${PREFIX}_reads.fasta" "$OUTDIR/${PREFIX}_reads.fasta" > "$OUTDIR/${PREFIX}.paf"
#Run miniasm
echo "Running miniasm..."
ml micromamba; micromamba activate miniasm;miniasm;miniasm -f "$OUTDIR/${PREFIX}_reads.fasta" "$OUTDIR/${PREFIX}.paf" -s 500 > "$OUTDIR/${PREFIX}.gfa"
#Convert GFA to FASTA
echo "Converting GFA to FASTA..."
awk '/^S/{print ">"$2"\n"$3}' "$OUTDIR/${PREFIX}.gfa" > "$OUTDIR/${PREFIX}.final.fasta"
echo "Assembly complete! Output files are in $OUTDIR"
</details>```
**Generate CANU 2.2 assemblies**
Note here that you will have to estimate your mitochondrial genome's size. Mine is too large, as I expected to get reads for the chloroplast genome as well.
```bash
for f in sample*fastq; do echo "ml micromamba; micromamba activate canu-env; canu -p mito -d "${f%.*}"CanuOut genomeSize=550k -nanopore-raw "$f ;done >canu.sh
Generate Raven assemblies
1
for f in sample*fastq; do echo "ml micromamba; micromamba activate raven;mkdir "${f%.*}"Raven ; cd "${f%.*}"Raven ; ln -s ../"$f"; raven "$f" >"${f%.*}"Assembly.fasta" ;done >raven.sh
Generate Unicycler Assemblies
1
2
3
4
5
6
# current location
/work/gif3/masonbrink/USDA/04_MitochondrialIsolationAndAssembly/Unicycler/
source bin/activate
# create assembly commands for each split
for f in *fastq; do echo "unicycler -l "$f" -t 36 --mode bold -o "${f%.*}"Unicycler --threads 36 --racon_path /work/gif3/masonbrink/Heuther/09_Unicycler/Unicycler/racon/build/bin/racon" ;done >UnicyclerRuns.sh
Evaluation of Assemblies
Acquire sapindaceae mitochondrial genes
Downloaded protein sequences from annotated Sapindus mukorossi mitochondrion
1
2
3
4
5
6
vi SmukorossiProteins.fasta
grep -c ">" SmukorossiProteins.fasta
41
#Rename proteins by protein name, not accession
awk '{if(NF>1) {print ">"$2} else {print $0}}' SmukorossiProteins.fasta >NamedSmukorossiProteins.fasta
Create a folder of uniquely named genome assemblies
Get all of the genomes into one folder with unique file names so that we can rerun the same scripts to evaluate them. This is not an essential step, but just makes the evaluation easier when you have so many different attempts. Here I use the “/” as a string separator to rename the files in a softlink. Note that I had three different read subsets, so I had to make sure the names were unique or the softlink would fail.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
#grabs all the unicycler assemblies and renames them by folder name, essentially by using the "/" as a separator to change the name.
for f in ../sample*Unicycler/assembly.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 2| awk '{print $1".fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >SoftlinkUnicycler.sh
sh SoftlinkUnicycler.sh
#grabs all flye assemblies and renames by folder name
for f in ../read_subsets/sample*FlyeOut/assembly.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk -F"\t" '{print $0".fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >SoftlinkFlye.sh
#softlinks all the miniasm assemblies which already have unique names
for f in ../read_subsets/*MiniasmOut_output/*final.fasta ; do ln -s $f;done
#softlinks and renames all canu assemblies
for f in ../read_subsets/*CanuOut/mito.contigs.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk -F"\t" '{print $0".fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >CanuSoftlink.sh
#softlinks and renames all Raven assemblies, which already have unique names
for f in ../read_subsets/*Raven/*Assembly.fasta ;do ln -s $f;done
#grabs all the unicycler assemblies and renames them by folder name
for f in ../read_subsetNuclearClean/*Unicycler/assembly.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk '{print $1"NuclearClean.fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >SoftlinkUnicycler2.sh
sh SoftlinkUnicycler2.sh
#grabs all flye assemblies and renames by folder name
for f in ../read_subsetNuclearClean/*FlyeOut/assembly.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk -F"\t" '{print $0"NuclearClean.fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >SoftlinkFlye2.sh
#softlinks all the miniasm assemblies which do not have unique names, and needed renamed this time
for f in ../read_subsetNuclearClean/*MiniasmOut_output/*final.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk -F"\t" '{print $0"NuclearClean.fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >SoftlinkMiniasm2.sh
#softlinks and renames all canu assemblies # all of these failed due to memory exhaustion
for f in ../read_subsetsNuclearClean/*CanuOut/mito.contigs.fasta; do echo "ln -s "$f ; echo $f |sed 's|/|\t|g' |cut -f 3| awk -F"\t" '{print $0"NuclearClean.fasta"}' ;done |tr "\n" " " |sed 's/ln -s/\nln -s/g' >CanuSoftlink2.sh
#softlinks and renames all Raven assemblies so they do not overwrite the ones above
for f in ../read_subsets/*Raven/*Assembly.fasta ;do ln -s $f ${f%.*}NuclearClean.fasta;done
Align mitochondrial proteins to assemblies
Align proteins to the genomes.
1
for f in *.fasta; do miniprot $f NamedSmukorossiProteins.fasta >${f%.*}.genes; done
How many of these 109 assemblies (all assemblies are not shown) had most of the genes on how many contigs and how large was the longest contig with mapping genes? Note I only printed assemblies with at least 35 unique mitochondrial gene alignments
1
for f in *genes; do paste <(ls -1 $f) <(cut -f 1 $f |sort|uniq|wc -l) <(cut -f 6 $f|sort|uniq|wc -l) <(cut -f 7 $f |sort -k1,1nr |awk '{print $1}' ) ;done |awk '$2>35' |less
| Assembly Name | Gene Count | Contig Count | Largest Contig |
|---|---|---|---|
| 1kClean60qualAssembly.genes | 38 | 3 | 339819 |
| 2kRawAssembly.genes | 38 | 1 | 536045 |
| 3kRawAssembly.genes | 38 | 1 | 536052 |
| 5kCleanAssembly.genes | 38 | 2 | 463238 |
| assembly.genes | 38 | 1 | 535684 |
| CleanReads3kAssembly.genes | 38 | 1 | 535684 |
| CleanReads5kAssembly.genes | 38 | 2 | 462995 |
| Flye5kRawAssembly.genes | 38 | 4 | 251593 |
| FlyeMiniprotAssembly.genes | 38 | 4 | 275765 |
| RawReads5kAssembly.genes | 37 | 2 | 372648 |
| sample_01FlyeOut.genes | 36 | 2 | 347764 |
| sample_02Assembly.genes | 39 | 4 | 470606 |
| sample_02FlyeOut.genes | 38 | 4 | 180048 |
| sample_03Assembly.genes | 39 | 3 | 463133 |
| sample_04Assembly.genes | 39 | 4 | 449804 |
| sample_04FlyeOut.genes | 38 | 5 | 257488 |
| sample_05Assembly.genes | 39 | 5 | 412423 |
| sample_05FlyeOut.genes | 37 | 4 | 205704 |
| sample_06Assembly.genes | 39 | 4 | 412399 |
| sample_06FlyeOut.genes | 37 | 3 | 225437 |
| sample_07Assembly.genes | 39 | 3 | 528267 |
| sample_07FlyeOut.genes | 38 | 2 | 381058 |
| sample_08Assembly.genes | 40 | 6 | 316564 |
| sample_08FlyeOut.genes | 38 | 5 | 123612 |
| sample_09Assembly.genes | 39 | 4 | 522057 |
| sample_09FlyeOut.genes | 38 | 5 | 229821 |
| sample_10Assembly.genes | 39 | 5 | 463042 |
| sample_10FlyeOut.genes | 36 | 4 | 128482 |
| sample_11Assembly.genes | 39 | 6 | 277342 |
| sample_11FlyeOut.genes | 38 | 6 | 163011 |
| sample_12Assembly.genes | 40 | 6 | 412414 |
| sample_12FlyeOut.genes | 36 | 5 | 252031 |
| UCManualInterventionRemoveNUMTReads.genes | 38 | 2 | 462795 |
So it looks like I commonly find 38-39 genes in these assemblies, but only two had the expected 40 genes. I ran a few assemblies with varying levels of filtering using all of the reads too, so those are included and some happen to be really good assemblies. The fewest contigs with the most genes seems to occur with assembly.fasta, which is an assembly using all trimmed reads that were able to get a 3kb alignment to the related Smukorossi mitochondrial genome. However, there are a couple assemblies that have all 40 genes, which we may use to incorporate into the singular contig at a later step.
Refining the best mitochondrial assembly
1
2
3
/work/gif3/masonbrink/USDA/04_MitochondrialIsolationAndAssembly/Unicycler/assemblies
#rename the assembly so I do not lose it
cp ../MitoNanopore3kRawUnicycler/assembly.fasta MitoNanopore3kRawUnicyclerassembly.fasta
This is a standard blast of my mitochondrial genome to itself.
1
2
makeblastdb -in MitoNanopore3kRawUnicyclerassembly.fasta -dbtype nucl
blastn -query MitoNanopore3kRawUnicyclerassembly.fasta -db MitoNanopore3kRawUnicyclerassembly.fasta -outfmt 6 -num_threads 36 -out MitoNanopore3kRawUnicyclerassembly.blastout
Blast Results: truncated to show the regions that are large and identical
| qseqid | sseqid | pident | length | mismatch | gapopen | qstart | qend | sstart | send | evalue | bitscore |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 1 | 100.000 | 535684 | 0 | 0 | 1 | 535684 | 1 | 535684 | 0.0 | 9.892e+05 |
| 1 | 1 | 99.851 | 43743 | 25 | 21 | 298590 | 342311 | 46555 | 2832 | 0.0 | 80382 |
| 1 | 1 | 99.851 | 43743 | 25 | 21 | 2832 | 46555 | 342311 | 298590 | 0.0 | 80382 |
| 1 | 1 | 99.969 | 31916 | 0 | 5 | 503771 | 535683 | 427604 | 459512 | 0.0 | 58874 |
| 1 | 1 | 99.969 | 31916 | 0 | 5 | 427604 | 459512 | 503771 | 535683 | 0.0 | 58874 |
| 1 | 1 | 99.984 | 6318 | 0 | 1 | 450654 | 456971 | 170106 | 163790 | 0.0 | 11660 |
| 1 | 1 | 99.984 | 6318 | 0 | 1 | 163790 | 170106 | 456971 | 450654 | 0.0 | 11660 |
| 1 | 1 | 99.968 | 6318 | 0 | 2 | 526823 | 533139 | 170106 | 163790 | 0.0 | 11655 |
| 1 | 1 | 99.968 | 6318 | 0 | 2 | 163790 | 170106 | 533139 | 526823 | 0.0 | 11655 |
| 1 | 1 | 99.640 | 2502 | 3 | 5 | 343015 | 345511 | 2501 | 1 | 0.0 | 4566 |
| 1 | 1 | 99.640 | 2502 | 3 | 5 | 1 | 2501 | 345511 | 343015 | 0.0 | 4566 |
So there are three duplicate regions that can be condensed to two regions. Since two are very close, I will just merge them
1-2501, 2832-46555, 503771-535683
Trim regions from the end of the assembly that are duplicated
Samtools is a pretty commonly used tool, so I will assume you have it available.
Indexes the fasta file
1
ml samtools;samtools faidx MitoNanopore3kRawUnicyclerassembly.fasta
These coordinates were used to extract the fasta from a blast to self identifying duplicated regions of the genome
1
echo "1:46555-503771" |samtools faidx MitoNanopore3kRawUnicyclerassembly.fasta -r - >TrimmedCleanMito.fasta
Check for the presence of all mitochondrial genes in the trimmed assembly
Map the proteins to your trimmed mitochondria
1
miniprot -S -j 0 TrimmedCleanMito.fasta NamedSmukorossiProteins.fasta >GenesInTrimmedCleanMito.paf
How many miniprot gene aligments are in this assembly out of the 38 it started with?
1
2
awk '{print $1}' GenesInTrimmedCleanMito.paf |sort|uniq|wc
38 38 205
Can we add the missing two genes that were found in other assemblies? This concatenates the names of the genes in my trimmed assembly with genes from a assembly with 40 miniprot gene alignments, finds the protein names that are only in one assembly, and extracts scaffolds with those protein alignments.
1
cat <(less ../assemblies/GenesInTrimmedCleanMito.paf |cut -f 1|sort|uniq ) <(cut -f 1 sample_08Assembly.genes |cut -f 1|sort|uniq) |sort|uniq -c |awk '$1==1{print $2}' |grep -w -f - sample_08Assembly.genes |cut -f 6 |cdbyank sample_08Assembly.fasta.cidx >MissingMitoScaffolds.fasta
Create a BLASTn database and run a standard self BLASTn
1
2
makeblastdb -in TrimmedCleanMito.fasta -dbtype nucl
blastn -db TrimmedCleanMito.fasta -query ../Assemblies/MissingMitoScaffolds.fasta -outfmt 6 -num_threads 8 -out MissingMitoScaffolds2TrimmedCleanMito.blastout
So there are two genes that were recognized as missing from my trimmed mitochondrial assembly
- orf119
- rps1
My first approach is to see if these contigs can align to my mitochondrial genome, if not then I evaluate the contigs with BLASTn to NCBI nucleotide database to see if they are more likely to be nuclear.
However, these genes map to two different 90kb scaffolds in both of the assemblies that were able to align these genes. Aligning these 90kb scaffolds to the TrimmedCleanMito.fasta assembly and the Sapindus mukrossi mitochondrial genome did not provide any hits, suggesting that these two genes may be NUMT genes integrated into the nuclear genome and can safely be ignored. If you suspect that you may have NUMTs in your assembly, then the best way to check this is to align your long reads to your mitochonrial genome (contaminated with NUMTs), and check to find large coverage gaps. Because mitochondrial reads should be at a much higher depth than nuclear NUMT reads, low coverage regions are likely to be NUMTs. This will allow you to find the NUMT regions in your assembly, and thus the NUMT reads, which you can remove with seqtk as above.
Circularize with circlator
This rotates the circular genome assembly so that it starts at a standard gene (dnaA), ensuring consistent start coordinates across species.
1
2
micromamba activate circlator
circlator fixstart TrimmedCleanMito.fasta ReOriented
Polish the mitochondrial genome
Align reads to this circularized mitochondrial assembly
Softlink the run script to run minimap pipeline to create a sorted bam. see below for script
1
ln -s ../../runMinimapNbamSort.sh
Use minimap to align reads to the genome to produce a sorted bam file.
1
sh runMinimapNbamSort.sh MitoNanopore3kRaw.fastq ReOriented.fasta
runMinimapNbamSort.sh
Install Pilon
Polish the mitochondria with Pilon to improve assembly quality
1
echo "ml pilon ; java -Xmx190g -Djava.io.tmpdir=TEMP -jar /opt/rit/el9/20230413/app/linux-rhel9-x86_64_v3/gcc-11.2.1/pilon-1.22-iojt4j5x62smcab6j4mjn77ejfqimebo/bin/pilon-1.22.jar --genome ReOriented.fasta --unpaired MitoNanopore3kRaw_ReOriented_minimap2_sorted.bam --outdir /work/gif3/masonbrink/USDA/04_MitochondrialIsolationAndAssembly/Unicycler/assemblies/ --changes --fix all --threads 36 ">pilon.sh
Results of assembly and polishing
The new_Assemblathon.pl script is freely available here: https://github.com/ISUgenomics/common_scripts/blob/master/new_Assemblathon.pl
1
new_Assemblathon.pl pilon.fasta
Results
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Number of scaffolds 1
Total size of scaffolds 457217
Longest scaffold 457217
Shortest scaffold 457217
Number of scaffolds > 1K nt 1 100.0%
Number of scaffolds > 10K nt 1 100.0%
Number of scaffolds > 100K nt 1 100.0%
Number of scaffolds > 1M nt 0 0.0%
Number of scaffolds > 10M nt 0 0.0%
Mean scaffold size 457217
Median scaffold size 457217
N50 scaffold length 457217
L50 scaffold count 1
n90 scaffold length 457217
L90 scaffold count 1
scaffold %A 27.60
scaffold %C 22.62
scaffold %G 22.30
scaffold %T 27.48
scaffold %N 0.00
scaffold %non-ACGTN 0.00
Number of scaffold non-ACGTN nt 0
Find the chloroplast genome within our assembled sequences
With all of the assemblies you we’ve made trying to assemble the mitochondria, we most likely assembled a chloroplast genome too, as they are much simpler than plant mitochondria. Let’s align the previously published chloroplast genome to the assemblies to see if we have some assembled. Note if you do not have a published version of your chloroplast assembly, you can use the same approach as what we did for the mitochondrial genome.
Find the chloroplast assembly
Download the chloroplast genome
1
NC_036099.1 Dodonaea viscosa chloroplast, complete genome -- 159,375bp
Map the published chloroplast assembly to our assemblies
1
for f in *fasta; do sh runMinimap.sh DviscosaChloroplast.fasta $f;done
Here we can see which contigs were of the appropriate size and complete via how much genome they aligned to
1
cat *minimap2.paf |awk '$7>150000 && $7<200000'|less
There were way too many, so filtered for even better assemblies. This will grab only those minimap alignments that are in the same orientation as the published assembly.
1
cat *minimap2.paf |awk '$4>157000 && $3<20'|less
In my case, I had 46 likely complete chloroplast assemblies, but only four were in the same orientation and started near the same base.
| Query Name | Query Length | Query Start | Query End | Strand | Target Name | Target Length | Target Start | Target End | Residue Matches | Alignment Block Length | Mapping Quality | tp | cm | s1 | s2 | dv | rl |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| NC_036099.1 | 159375 | 5 | 159366 | - | tig00000001 | 194930 | 8636 | 167738 | 139848 | 159617 | 60 | tp:A:P | cm:i:25896 | s1:i:139710 | s2:i:80342 | dv:f:0.0081 | rl:i:0 |
| NC_036099.1 | 159375 | 5 | 158489 | + | Utg177960 | 268827 | 56996 | 215246 | 138870 | 158751 | 60 | tp:A:P | cm:i:25672 | s1:i:138733 | s2:i:108725 | dv:f:0.0090 | rl:i:962 |
| NC_036099.1 | 159375 | 5 | 159366 | + | tig00000001 | 190469 | 30999 | 190097 | 139817 | 159615 | 60 | tp:A:P | cm:i:25895 | s1:i:139681 | s2:i:72177 | dv:f:0.0081 | rl:i:0 |
| NC_036099.1 | 159375 | 5 | 159366 | - | Utg177424 | 252033 | 5114 | 147499 | 140664 | 159651 | 60 | tp:A:P | cm:i:25994 | s1:i:138502 | s2:i:71811 | dv:f:0.0086 | rl:i:960 |
Column 7 tells us the length of the contig that the chloroplast genome aligned to, and it appears that either this genome is much larger than the published version or it contains a false amount of duplication. The easiest way to identify if the contig is too big then perform a self-blastn on the contig.
BLASTn of Utg177960 to Utg177960
Index and extract the chloroplast contig for BLAST
1
2
cdbfasta sample_06Assembly.fasta
echo "Utg177960" |cdbyank sample_06Assembly.fasta.cidx >MyChloroplast.fasta
Create the blast database
1
makeblastdb -in MyChloroplast.fasta -dbtype nucl
Run the self blast
1
blastn -db MyChloroplast.fasta -query MyChloroplast.fasta -outfmt 6 -out Mychloroplast2mychloroplast.blastout
How much of the assembly is identical? The top line will always be the whole assembly, but the next best hits which are reciprocal best hits that will tell us the positions of duplication.
| qseqid | sseqid | pident | length | mismatch | gapopen | qstart | qend | sstart | send | evalue | bitscore |
|————|————|———|——–|———-|———|———|———|———|———|——–|———–|
| Utg177960 | Utg177960 | 100.000 | 268827 | 0 | 0 | 1 | 268827 | 1 | 268827 | 0.0 | 4.964e+05 |
| Utg177960 | Utg177960 | 99.980 | 56113 | 0 | 7 | 159146 | 215251 | 1 | 56109 | 0.0 | 1.036e+05 |
| Utg177960 | Utg177960 | 99.980 | 56113 | 0 | 7 | 1 | 56109 | 159146 | 215251 | 0.0 | 1.036e+05 |
| Utg177960 | Utg177960 | 99.983 | 52625 | 1 | 7 | 215252 | 267872 | 144613 | 197233 | 0.0 | 97123 |
| Utg177960 | Utg177960 | 99.983 | 52625 | 1 | 7 | 144613 | 197233 | 215252 | 267872 | 0.0 | 97123 |
So there are two duplicate regions from 1-56109 and 215252-267872, lets extract the remaining sequence.
1
echo "Utg177960 56109 215252" |cdbyank sample_06Assembly.fasta.cidx -R |bioawk -c fastx '{print $name,length($seq)}' >UnpolishedMyChloroplastDeduplicated.fasta
Circularize with circlator
This rotates the circular genome assembly so that it starts at a standard gene (dnaA), ensuring consistent start coordinates across species.
1
2
ml micromamba; micromamba activate circlator
circlator fixstart UnpolishedMyChloroplastDeduplicated.fasta MyReOrientedChloroplastGenome
Polish chloroplast assembly
Softlink the Minimap to sorted bam script see above
1
ln -s ../runMinimapNbamSort.sh
Align the reads, sort and index
1
sh runMinimapNbamSort.sh MitoNanopore3kRaw.fastq ReOrientedChloroplastGenome.fasta
Polish the chloroplast genome
1
echo " ml pilon ; java -Xmx190g -Djava.io.tmpdir=TEMP -jar /opt/rit/el9/20230413/app/linux-rhel9-x86_64_v3/gcc-11.2.1/pilon-1.22-iojt4j5x62smcab6j4mjn77ejfqimebo/bin/pilon-1.22.jar --genome MyReOrientedChloroplastGenome.fasta --unpaired MitoNanopore3kRaw_ReOrientedChloro_minimap2_sorted.bam --outdir /work/gif3/masonbrink/USDA/04_MitochondrialIsolationAndAssembly/Unicycler/ --changes --fix all --threads 36 ">pilonAllChloro.sh
Results of assembly and polishing
The new_Assemblathon.pl script is freely available here: https://github.com/ISUgenomics/common_scripts/blob/master/new_Assemblathon.pl
1
new_Assemblathon.pl MyReOrientedChloroplastGenomePilon.fasta
Results
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
---------------- Information for assembly 'MyReOrientedChloroplastGenome.fasta' ----------------
Number of scaffolds 1
Total size of scaffolds 159144
Longest scaffold 159144
Shortest scaffold 159144
Number of scaffolds > 1K nt 1 100.0%
Number of scaffolds > 10K nt 1 100.0%
Number of scaffolds > 100K nt 1 100.0%
Number of scaffolds > 1M nt 0 0.0%
Number of scaffolds > 10M nt 0 0.0%
Mean scaffold size 159144
Median scaffold size 159144
N50 scaffold length 159144
L50 scaffold count 1
n90 scaffold length 159144
L90 scaffold count 1
scaffold %A 30.65
scaffold %C 19.17
scaffold %G 18.73
scaffold %T 31.45
scaffold %N 0.00
scaffold %non-ACGTN 0.00
Number of scaffold non-ACGTN nt 0
Conclusion
This tutorial has outlined a comprehensive workflow for assembling mitochondrial and chloroplast genomes with nuclear-targeted sequencing reads. Researchers can now effectively isolate, assemble, and refine organellar genomes even if they were not the focus of the original research. Key steps include the selection of appropriate assemblers based on read types, deduplication of circular assemblies, circularization of the genome, and polishing to correct errors and improve accuracy. This tutorial facilitates the generation of high-quality mitochondrial genome assemblies, enabling better downstream comparative genomics analyses.
Software Citations and Resources
| Tool | Citation | GitHub Repository |
|---|---|---|
| Unicycler | Wick RR, Judd LM, Gorrie CL, Holt KE. Unicycler: Resolving bacterial genome assemblies from short and long sequencing reads. PLoS Comput Biol. 2017;13(6):e1005595. doi:10.1371/journal.pcbi.1005595 | Unicycler GitHub |
| Miniasm | Li H. Minimap and miniasm: fast mapping and de novo assembly for noisy long sequences. Bioinformatics. 2016;32(14):2103–2110. doi:10.1093/bioinformatics/btw152 | Miniasm GitHub |
| Flye | Kolmogorov M, Yuan J, Lin Y, Pevzner PA. Assembly of long, error-prone reads using repeat graphs. Nat Biotechnol. 2019;37(5):540–546. doi:10.1038/s41587-019-0072-8 | Flye GitHub |
| Canu | Koren S, Walenz BP, Berlin K, Miller JR, Phillippy AM. Canu: scalable and accurate long-read assembly via adaptive k-mer weighting and repeat separation. Genome Res. 2017;27(5):722–736. doi:10.1101/gr.215087.116 | Canu GitHub |
| Raven | Vaser R, Šikić M. Raven: a de novo genome assembler for long reads. bioRxiv. 2020. doi:10.1101/2020.08.07.242461 | Raven GitHub |
| Circlator | Hunt M, Silva ND, Otto TD, Parkhill J, Keane JA, Harris SR. Circlator: automated circularization of genome assemblies using long sequencing reads. Genome Biol. 2015;16:294. doi:10.1186/s13059-015-0849-0 | Circlator GitHub |
| Pilon | Walker BJ, Abeel T, Shea T, et al. Pilon: an integrated tool for comprehensive microbial variant detection and genome assembly improvement. PLoS One. 2014;9(11):e112963. doi:10.1371/journal.pone.0112963 | Pilon GitHub |
Additional Resources
I purposefully made this tutorial with generic assemblers so that specific software is not needed for organelle assembly. However, there are tools specifically designed for mitochondrial genome assembly.
- MitoZ: https://github.com/linzhi2013/MitoZ
- GetOrganelle: https://github.com/Kinggerm/GetOrganelle
- NOVOPlasty: https://github.com/ndierckx/NOVOPlasty