Title: Phylogenetic Authentication of Amplicon Sequence Variants in Single-Specimen Metabarcoding of Tropical Insects

Authors: Sarawut Ounjai, Huaxi Liu, Zichen Zhou, Maria Pestana Correia, Thomas J. Creedy, Carmelo Andújar, Paula Arribas, Alfried P. Vogler

Journal: Molecular Ecology Resources (in press, 2026)

Dryad DOI: https://doi.org/10.5061/dryad.v41ns1s9q

GitHub (analysis scripts): https://github.com/OunjaiS/ASV-Selection-Taxonomic-Assignment

Zenodo (archived scripts): https://doi.org/10.5281/zenodo.18732406

Total dataset size: 1.3 GB

---

Overview

This dataset contains COI amplicon sequence data from 18,533 Coleoptera (beetle) specimens collected across 14 countries and 119 families, generated by individual Illumina MiSeq barcoding. Each specimen's ASVs have been classified into five categories using an integrated authentication framework combining abundance filtering and phylogenetic placement. Any empty cells represent missing data.

---

Files in This Dataset

Raw Data

1. ASV_Authentication_Results_030925.csv (75.89 MB)

Main dataset. One row per ASV–specimen combination (175,954 rows). Contains all classification results, read counts, phylogenetic distances, and specimen metadata.

Key columns:

Classification values for asv_classification:

• Authenticated — passes MRCT ≥ 4 AND phylogenetic match to morphological family
• Technical_Artifacts — fails MRCT or chimeric; typically 1–3 reads
• Environmental_Contamination — non-Coleoptera phylogenetic placement
• Nuclear_Pseudogenes — intra-individual variant at low abundance; close phylogenetic distance
• Failed — low read count; insufficient data for classification

---

2. ASV_sequences_64544.fasta (28.13 MB)

FASTA file of 64,544 unique ASV sequences after removal of sequences with in-frame stop codons or frameshifts under NCBI Genetic Code Table 5 (Invertebrate Mitochondrial). Sequence IDs match ASV_ID in the main CSV. Used as input for Scripts 2, 4, and 6.

---

3. Tree_fasttree_bt_supermatrix_C13K_MMG_COX1_rooted_RN.tree (4.38 MB)

Newick-format maximum-likelihood phylogenetic reference tree constructed from 13,380 COI mitogenome sequences using FastTree v2.1.11 (GTR+CAT approximation model). Rooted and tip-labelled. Used for phylogenetic placement of all 175,954 ASVs in the authentication framework (Script 2). Visualised in Supplementary Figure S1.

---

4. lmm_readcount_data.csv (7.7 MB)

Read count data for the linear mixed model comparing authenticated vs. non-authenticated ASVs (Script 5, Analysis 1). One row per ASV–specimen record (175,954 rows).

---

Analysis Outputs

5. ASV_Complete_Analysis.csv (505.92 MB)

Output of Script 1 (1_sequence_qc.ipynb). QC metrics for all ASVs including ORF detection across 6 reading frames, GC content, codon usage, and frameshift status under NCBI Genetic Code Table 5. Primary input for Script 2.

---

6. ASV_Corrected_Sequences.fasta (86.52 MB)

Output of Script 1. Frameshift-corrected ASV sequences under NCBI Genetic Code Table 5.

---

7. ASV_Final_Classification.csv (535.67 MB)

Output of Script 2 (2_asv_classification.ipynb). Complete ASV records with rule-based and Random Forest classification features appended (175,954 rows). Primary input for Script 3.

---

8. ASV_Sequence_Summary.csv (51.93 MB)

Output of Script 2. Per-sequence summary statistics and classification feature values used by the Random Forest classifier.

---

9. Classification_Model.pkl (6.53 MB)

Trained Random Forest classifier (scikit-learn v1.3.0, Python 3.11) from Script 2. Can be loaded with Python pickle to classify new ASV data without retraining. Requires scikit-learn ≥ 1.1.

---

Supplementary Tables

10. Table_S1_Country_Success.csv (1.32 KB)

Authentication success rates by country (14 countries). Columns: country, total specimens, authenticated specimens, success rate (%).

11. Table_S2_Method_Success.csv (1.14 KB)

Authentication success rates by collection method (pitfall trap, Malaise trap, active collection, etc.).

12. Table_S3_Batch_Success.csv (862 B)

Authentication success rates by sequencing batch (10 Illumina MiSeq runs).

13. Table_S4_All_Families.csv (8.98 KB)

Authentication success rates for all 119 beetle families.

14. Table_S5_Statistical_Tests.csv (657 B)

Chi-square and Mann-Whitney U test results for factors affecting authentication success (country, method, batch, family, specimen size class).

15. Table_S6_DADA2_analysis.csv (2.91 KB)

Conceptual framework comparison between the VSEARCH/UNOISE3 pipeline and DADA2, including authentication gap analysis (57,970 biological sequences that pass denoising but require authentication).

16. Table_S7_RSCU_comparison.csv (6.97 KB)

Relative Synonymous Codon Usage (RSCU) comparison between matched authenticated anchor ASVs (n = 330) and confirmed NUMTs (n = 930) for all 62 sense codons under NCBI Genetic Code Table 5. Includes Benjamini-Hochberg adjusted p-values; 12 codons differ significantly (BH-FDR < 0.05).

17. Table_S8_cross-specimen_matched_comparison.csv (884 B)

Within-specimen linear mixed model (LMM) results comparing GC content, GC at each codon position, and Effective Number of Codons (ENC) between confirmed NUMTs and matched anchor ASVs. Includes ICC values (0.896–0.949) and cross-specimen Mann-Whitney U test results.

---

Supplementary Figures

18. Supplementary_Figure_S1_mitogenome_reference_tree.pdf (476.17 KB)

Maximum-likelihood phylogenetic reference tree constructed from 13,380 COI mitogenome sequences (FastTree v2.1.11, GTR+CAT). Used for phylogenetic placement of all 175,954 ASVs.

19. Supplementary_Figure_S2.pdf (338.52 KB)

DADA2 benchmarking comparison figure. Illustrates the authentication gap — sequences passing denoising quality filters that require specimen-level authentication to classify correctly.

---

Supplementary Materials

20. Supplementary_Material_1_Primers.xlsx (12.68 KB)

PCR primer sequences, tags, and library preparation details for all sequencing batches.

21. Supplementary_Material_2_Bioinformatics_Pipeline.md (15.69 KB)

Step-by-step documentation of the bioinformatics pipeline from raw FASTQ files to authenticated ASVs, including all software commands, parameters, and version numbers.

22. Supplementary_Material_3_mitogenomes.xlsx (1.11 MB)

List of 13,380 COI mitogenome sequences used to construct the phylogenetic reference tree, including GenBank accession numbers, taxonomy, and geographic origin.

23. Supplementary_Material_4_Phylogenetic_Framework.md (17.01 KB)

Documentation of the phylogenetic framework construction and the three reference tree variants used to assess the influence of taxon sampling density on authentication.

---

How to Use This Dataset

Reproduce all analyses from scratch.

1. Download all raw data files (files 1–4) and place them according to the GitHub README
2. Run Scripts 1–6 in numbered order

Skip to statistical analyses (Scripts 3–6)

1. Download ASV_Final_Classification.csv (file 7), ASV_sequences_64544.fasta (file 2), and lmm_readcount_data.csv (file 4)
2. Place ASV_Final_Classification.csv in output/2_classification_analysis/
3. Run Scripts 3–6

Use the authenticated ASVs only

Filter ASV_Authentication_Results_030925.csv for rows where asv_classification == "Authenticated". This yields 15,901 records representing 12,922 unique sequences from 14,715 specimens.

import pandas as pd
df = pd.read_csv('ASV_Authentication_Results_030925.csv', low_memory=False)
authenticated = df[df['asv_classification'] == 'Authenticated']

---

Software Requirements

Full installation instructions: see GitHub repository README.

---

Sampling Locations

Collection methods: pitfall traps, Malaise traps, leaf litter extraction, active hand collection, light trapping.
Preservation: ethanol or SDS+EDTA buffer.

---

Genetic Code

All sequence analyses use NCBI Genetic Code Table 5 (Invertebrate Mitochondrial):

• TGA → Trp (W)
• ATA → Met (M)
• AGA, AGG → Ser (S)
• TAA, TAG → Stop

This code yields 62 sense codons (not 61 as in the standard code).

---

License

CC0 1.0 Universal — no rights reserved. Data may be freely used, shared, and adapted without restriction.

---

Citation

If you use this dataset, please cite:
Ounjai, S., Liu, H., Zhou, Z., Pestana Correia, M., Creedy, T.J., Andújar, C., Arribas, P., and Vogler, A.P. (in press). Diversity and Phylogenetic Placement of Amplicon Sequence Variants in Single-Specimen Metabarcoding of Tropical Insects. Molecular Ecology Resources.

Dataset: https://doi.org/10.5061/dryad.v41ns1s9q