- What is ancseq?
- Installation
- Usage
- Outputs
- Workflow in ancseq
- How does ancseq calculate codon probabilities in DNA mode?
Ancestral sequence reconstruction is a technique to reconstruct ancestral states from a multiple sequence alignment. ancseq is a wrapper tool to reconstruct ancestral sequences using IQ-TREE. See more detail workflow in ancseq here.
Sugihara Y, Kourelis J, Contreras MP, Pai H, Harant A. Selvaraj M, Toghani A, Martinez-Anaya C*, Kamoun S* (2025) Helper NLR immune protein NRC3 evolved to evade inhibition by a cyst nematode virulence effector. PLOS Genetics, 21:e1011653 *Corresponding authors
- IQ-TREE (must be on PATH)
You can install ancseq with the dependencies using anaconda.
git clone https://github.com/YuSugihara/ancseq.git
cd ancseq
conda env create -f ancseq.yml
conda activate ancseq
pip install .Note: IQ-TREE is installed via ancseq.yml, so no separate installation is needed. After activating the environment you can verify with:
iqtree -h
If you install ancseq outside this conda environment, install IQ-TREE separately and ensure its executable is on PATH.
$ ancseq -h
usage: ancseq -s <ALIGNED_FASTA> -m <MODE> -o <OUT_DIR> [-t <INT>]
ancseq version 1.3.0
options:
-h, --help show this help message and exit
-s, --seq Sequence alignment in FASTA format.
-m, --mode Sequence type. [DNA/AA/CODON]
-o, --out Output directory. The given name must not exist.
-t, --threads Number of threads. [4]
-b, --bootstrap Replicate for bootstrap. [1000]
--max-report Maximum number of ambiguous sites to report at the same position. [5]
--min-prob Minimum probability of being reported as an ambiguous site. [0.05]
--min-gap-prob Minimum probability of replacing the ancestral state with a gap. [0.5]
--fast Use -fast option in IQ-TREE [FALSE]
--model Specify substitution model for IQ-TREE. IQ-TREE searches the best substitution
model using ModelFinder by default [MFP]
--outgroup Specify outgroup for IQ-TREE. [None]
--stop-codon-prob Stop calculation of codon probabilities in DNA mode [FALSE]
--asr-only Skip building tree and reconstruct ancestral states only [FALSE]
-v, --version show program's version number and exit- Example 1 : Running ancseq for nucleotide sequence alignment
- Example 2 : Running ancseq for amino acid sequence alignment
- Example 3 : Running ancseq for codon sequence alignment
- Example 4 : Running ancseq specifying outgroup
- Example 5 : Running ancseq with --fast option
ancseq -s test_nuc.fasta \
-m DNA \
-o out_dir
-s : Nucleotide sequence alignment in fasta format.
-m : Sequence type.
-o : Name of the output directory. The given name should not exist.
ancseq -s test_nuc.fasta \
-m AA \
-o out_dir-s : Amino acid sequence alignment in fasta format.
-m : Sequence type.
-o : Name of the output directory. The given name should not exist.
ancseq -s test_codon.fasta \
-m CODON \
-o out_dir-s : Codon sequence alignment in fasta format.
-m : Sequence type.
-o : Name of the output directory. The given name should not exist.
You can reconstruct the ancestral states without specifying an outgroup. However, the ancestral states of a node may be misinterpreted when you visualize the tree. Therefore, we recommend specifying an outgroup to avoid misinterpretation. IQ-TREE converts the rooted tree to an unrooted tree by default.
ancseq -s test_nuc.fasta \
-m DNA \
--outgroup seq_id \
-o out_dir
-s : Nucleotide sequence alignment in fasta format.
-m : Sequence type.
--outgroup : Sequence ID of outgroup.
-o : Name of the output directory. The given name should not exist.
ancseq -s test_nuc.fasta \
-m DNA \
-o out_dir \
--fast-s : Nucleotide sequence alignment in fasta format.
-m : Sequence type.
-o : Name of the output directory. The given name should not exist.
--fast : Fast search to resemble FastTree (IQ-TREE -fast).
Inside of OUT_DIR is like below.
├── 00_tree
│ ├── 00_iqtree.err
│ ├── 00_iqtree.out
│ ├── test_nuc.fasta
│ ├── test_nuc.fasta.bionj
│ ├── test_nuc.fasta.ckp.gz
│ ├── test_nuc.fasta.contree
│ ├── test_nuc.fasta.iqtree
│ ├── test_nuc.fasta.log
│ ├── test_nuc.fasta.mldist
│ ├── test_nuc.fasta.model.gz
│ ├── test_nuc.fasta.splits.nex
│ └── test_nuc.fasta.treefile
├── 10_asr
│ ├── 10_iqtree.err
│ ├── 10_iqtree.out
│ ├── test_nuc.fasta
│ ├── test_nuc.fasta.ckp.gz
│ ├── test_nuc.fasta.iqtree
│ ├── test_nuc.fasta.log
│ ├── test_nuc.fasta.state.gz
│ └── test_nuc.fasta.treefile
├── 20_indels
│ ├── 20_iqtree.err
│ ├── 20_iqtree.out
│ ├── test_nuc.fasta.binary
│ ├── test_nuc.fasta.binary.ckp.gz
│ ├── test_nuc.fasta.binary.iqtree
│ ├── test_nuc.fasta.binary.log
│ ├── test_nuc.fasta.binary.state.gz
│ └── test_nuc.fasta.binary.treefile
└── 30_result
├── ancestral_state_result.treefile
├── ancestral_state_result.fasta
├── ancestral_state_result_with_gap.fasta
├── ancestral_state_result.sort.tsv
├── ancestral_state_result.tsv.gz
└── ancestral_state_result.codon_prob.tsv.gz
- The phylogenetic tree reconstructed by IQ-TREE can be found in
00_tree. - The results of the ancestral sequence reconstruction can be found in
30_result.ancestral_state_result.treefile: Phylogenetic tree with the node labels.ancestral_state_result.fasta: FASTA file of the ancestral sequences without gaps.ancestral_state_result_with_gap.fasta: FASTA file of the ancestral sequences with gaps.ancestral_state_result.sort.tsv: Probabilities of the ancestral states.
- IQ-TREE command 1 : Building a phylogenetic tree.
- IQ-TREE command 2 : Reconstructing ancestral sequences.
- IQ-TREE command 3 : Reconstructing insertions and deletions (INDELs).
iqtree -s ${INPUT_FASTA} \
-st ${SEQ_TYPE} \
-T ${NUM_THREADS} \
-B ${NUM_BOOTSTRAP} \
-m MFP \
1> /OUT_DIR/00_tree/00_iqtree.out \
2> /OUT_DIR/00_tree/00_iqtree.err-s : Sequence alignment in fasta format.
-st : Sequence type.
-T : Number of threads.
-B : Replicates for ultrafast bootstrap.
-m MFP : Extended model selection followed by tree inference.
If you specify the --fast option in ancseq, the IQ-TREE command 1 will change as follows.
iqtree -s ${INPUT_FASTA} \
-st ${SEQ_TYPE} \
-T ${NUM_THREADS} \
--alrt ${NUM_BOOTSTRAP} \
-m MFP \
--fast \
1> /OUT_DIR/00_tree/00_iqtree.out \
2> /OUT_DIR/00_tree/00_iqtree.err-s : Sequence alignment in fasta format.
-st : Sequence type.
-T : Number of threads.
--alrt : Replicates for SH approximate likelihood ratio test.
-m MFP : Extended model selection followed by tree inference.
--fast : Fast search to resemble FastTree.
iqtree -asr \
-s ${INPUT_FASTA} \
-te /OUT_DIR/00_tree/${INPUT_FASTA}.treefile \
-st ${SEQ_TYPE} \
-T ${NUM_THREADS} \
-m ${MODEL} \
-o ${OUTGROUP} \
-keep_empty_seq \
1> /OUT_DIR/10_asr/10_iqtree.out \
2> /OUT_DIR/10_asr/10_iqtree.err-asr : Ancestral state reconstruction by empirical Bayes.
-s : Sequence alignment in fasta format.
-te : Tree file.
-st : Sequence type.
-T : Number of threads.
-m : Model name.
-o : Outgroup sequence ID (only present if --outgroup was specified).
-keep_empty_seq : Keep empty sequences in the alignment.
iqtree -asr \
-s ${INPUT_FASTA}.binary \
-te /OUT_DIR/00_tree/${INPUT_FASTA}.treefile \
-st BIN \
-T ${NUM_THREADS} \
-blfix \
-m JC2 \
-o ${OUTGROUP} \
-keep_empty_seq \
1> /OUT_DIR/20_indels/20_iqtree.out \
2> /OUT_DIR/20_indels/20_iqtree.err-asr : Ancestral state reconstruction by empirical Bayes.
-s : Sequence alignment in fasta format.
-te : Tree file.
-st BIN : Binary sequence type.
-T : Number of threads.
-blfix : Fix branch lengths of tree passed via -t or -te.
-m JC2 : Jukes-Cantor type binary model.
-o : Sequence ID of the outgroup if you specify the outgroup with --outgroup.
-keep_empty_seq : Keep empty sequences in the alignment.
-
Aadland K, Pugh C, Kolaczkowski B. 2019. High-Throughput Reconstruction of Ancestral Protein Sequence, Structure, and Molecular Function. In: Sikosek T ed. Computational Methods in Protein Evolution. Methods in Molecular Biology. New York, NY: Springer, 135–170. DOI: 10.1007/978-1-4939-8736-8_8.
-
VanAntwerp J, Finneran P, Dolgikh B, Woldring D. 2022. Ancestral Sequence Reconstruction and Alternate Amino Acid States Guide Protein Library Design for Directed Evolution. In: Traxlmayr MW ed. Yeast Surface Display. Methods in Molecular Biology. New York, NY: Springer US, 75–86. DOI: 10.1007/978-1-0716-2285-8_4.
In DNA mode, ancseq approximates codon probabilities by multiplying the nucleotide posterior probabilities of the three positions (independence assumption). This is useful when the full IQ-TREE codon model would be too slow.
For codon index j (1-based), the three nucleotide alignment columns are:
j1 = 3*(j-1)+1
j2 = 3*(j-1)+2
j3 = 3*(j-1)+3
Let p_{j1,X}, p_{j2,Y}, p_{j3,Z} be the posterior probabilities of nucleotides X, Y, Z at those three positions.
For any codon xyz:
P_j(xyz) = p_{j1,x} * p_{j2,y} * p_{j3,z}
Example (methionine, ATG):
P_j(ATG) = p_{j1,A} * p_{j2,T} * p_{j3,G}
This ignores positional dependence modeled in full codon substitution analyses but is fast and practical.