MAG Quality and Annotation - iffatAGheyas/bioinformatics-tutorial-wiki GitHub Wiki

After recovering metagenome-assembled genomes (MAGs), it’s essential to assess their quality (completeness, contamination) and then annotate gene content. We cover three key tools: CheckM, BUSCO, and Prokka.

# Using conda (recommended)
conda install -c bioconda checkm-genome busco prokka

You can also install Prokka via sudo apt install prokka on Debian/Ubuntu, but conda ensures version consistency.

CheckM uses lineage-specific marker sets to estimate genome completeness and contamination.

# 1) Create an output directory
mkdir -p quality/checkm

# 2) Run the lineage workflow on your MAG folder
checkm lineage_wf \
  --reduced_tree \
  -x fa \
  bins/DASTool_DASTool_bins/ \
  quality/checkm \
  -t 8

• x fa specifies your MAG files have extension .fa

• reduced_tree speeds up tree placement for prokaryotes

• lineage.ms (marker set used)

• storage/final_summary.tsv summary table with columns:

  • Bin Id

  • Completeness (%)

  • Contamination (%)

  • Strain heterogeneity (%)

Example final_summary.tsv:

• Goal: MAGs with ≥ 90 % completeness and ≤ 5 % contamination are considered high-quality.

BUSCO searches for conserved single-copy orthologous genes to assess completeness and fragmentation.

# 1) Create output directory
mkdir -p quality/busco

# 2) Run BUSCO on each MAG
for bin in bins/DASTool_DASTool_bins/*.fa; do
  name=$(basename "$bin" .fa)
  busco \
    -i "$bin" \
    -l bacteria_odb10 \       # bacterial lineage dataset
    -o "busco/$name" \
    -m genome \
    -c 8
done

• l bacteria_odb10 uses the bacteria-specific BUSCO set (1 000+ orthologs).

• BUSCO will download the lineage data automatically.

• short_summary..txt

• run_/full_table.tsv

C:96.4%[S:94.1%,D:2.3%],F:1.2%,M:2.4%,n:1240

• C = Complete BUSCOs (S = single copy, D = duplicated)

• F = Fragmented BUSCOs

• M = Missing BUSCOs

• n = Number of BUSCO groups searched

Prokka predicts coding sequences (CDS), rRNAs, tRNAs, and other features, assigning functional annotations.

# 1) Create annotation output dir
mkdir -p annotation/prokka

# 2) Annotate each MAG
for bin in bins/DASTool_DASTool_bins/*.fa; do
  name=$(basename "$bin" .fa)
  prokka \
    --outdir annotation/prokka/"$name" \
    --prefix "$name" \
    --cpus 8 \
    --kingdom Bacteria \
    "$bin"
done

• .gff – GFF3 with gene features

• .gbk – GenBank file

• .faa – protein translations (FASTA)

• .ffn – nucleotide CDS (FASTA)

• .tbl – feature table for GenBank

• .txt – summary statistics (number of CDS, rRNA, tRNA)

# Assembly: bin.1.fa
# Contigs: 52  (total length: 3,745,128 bp)
# Genes:   3,650
# CDS:     3,550
# rRNA:    3
# tRNA:    45

quality/
├─ checkm/
│   ├─ lineage.ms
│   └─ storage/final_summary.tsv
└─ busco/
    ├─ bin.1/
    │   ├─ short_summary.bin.1.txt
    │   └─ run_bin.1/full_table.tsv
    └─ bin.2/ …

annotation/
└─ prokka/
    ├─ bin.1/
    │   ├─ bin.1.gff
    │   ├─ bin.1.faa
    │   ├─ bin.1.ffn
    │   └─ …
    └─ bin.2/ …

• Use CheckM and BUSCO together for complementary completeness estimates.

• Focus downstream analyses on high-quality MAGs (≥ 90 % complete, ≤ 5 % contaminated).

• Retain Prokka annotations (GFF and protein FASTA) for metabolic reconstruction and comparative genomics.