Date: 08/08/2025
Authors: Arthur Monjot
Dependencies: R; ITSx; vsearch; cutadapt; seqkit
gunzip rawdata/OBAMA01_Boxes-4_FullSequences_with-OTU-IDs.fasta.gz
mkdir -p Outputs/LongReads_analysis/region/
ITSx -i rawdata/OBAMA01_Boxes-4_FullSequences_with-OTU-IDs.fasta -o Outputs/LongReads_analysis/region/Thio_LR_extracted --preserve T --cpu 12 --save_regions all
cat Outputs/LongReads_analysis/region/Thio_LR_extracted.SSU.fasta | grep "^>" | wc -l
7055 SSU derived from LR sequences remain. The result appears convenient
The SR reverse primer: TAReukREV3
5"-ACTTTCGTTCTTGATYRA-3"
reverse complement: TYRATCAAGAACGAAAGT
"T[CT][AG]ATCAAGAACGAAAGT"
nohup cutadapt -a TYRATCAAGAACGAAAGT";min_overlap=18" -o Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.fasta --action retain -e 0.2 --discard-untrimmed Outputs/LongReads_analysis/region/Thio_LR_extracted.SSU.fasta > Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.log
cat Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.fasta | grep "^>" | wc -l
6841 sequences remain after the selection with cutadapt. This is pretty better than the solution using regular expression (i.e. 5725).
We use the seqkit tool:
seqkit stats Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.fasta > Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.stat
file | format | type | num_seqs | sum_len | min_len | avg_len | max_len Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.fasta | FASTA | DNA | 6,841 | 2,642,458 | 236 | 386.3 | 555
We have a csv table with in column 1 the header and in column 2 the sequence. We have to transform this file in fasta file.
cat rawdata/database_18SV4_Thio.csv | awk -F";" '{print ">"$1"\n"$2}' | tail -n+3 > Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta
cat Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta | grep ">" | wc -l
They are 25250 SR OTUs sequences.
The LR forward primer: Euk575Fngs
5"-ASCYGYGGTAAYWCCAGC-3"
"A[CG]C[CT]G[CT]GGTAA[CT][AT]CCAGC"
The SR forward primer: TAReukFWD1
5"-"CCAGCASCYGCGGTAAT-3"
"CCAGCA[CG]C[CT]GCGGTAAT""
Since bot primers are very closed in their position on the SSU, we choose to not trim the SR
#nohup cutadapt -g ASCYGYGGTAAYWCCAGC -o Outputs/LongReads_analysis/region/Thio_SR_cutadapt_trimmed.V4.fasta --action retain -e 0.2 Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta > Outputs/LongReads_analysis/region/Thio_SR_OTUs_cutadapt_trimmed.V4.log
We use seqkit tool:
seqkit stats Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta > Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.stat
file | format | type | num_seqs | sum_len | min_len | avg_len | max_len Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta | FASTA | DNA | 25,250 | 8,573,229 | 232 | 339.5 | 451
We use Vsearch to align all LR and SR-derived region. We check, in the same time, if the query and subject sequence length is up to 80% in coverage.
mkdir Outputs/LongReads_analysis/Clustering/
cat Outputs/LongReads_analysis/region/Thio_SR_OTUs.V4.fasta Outputs/LongReads_analysis/region/Thio_LR_cutadapt_trimmed.V4.fasta > Outputs/LongReads_analysis/Clustering/All_V4_sequences.fasta
vsearch --allpairs_global Outputs/LongReads_analysis/Clustering/All_V4_sequences.fasta --id 0.8 --query_cov 0.8 --target_cov 0.8 --threads 12 --blast6out Outputs/LongReads_analysis/Clustering/All_V4_allvsall_cover80.tsv
It takes a while!