APPLES stands for Accurate Phylogenetic Placement with LEast Squares and addresses the problem of phylogenetic placement of DNA and protein sequences into an already existing reference tree. APPLES is a command-line tool and it can run on Linux, Mac OSX, and Windows.
- Metin Balaban, Shahab Sarmashghi, and Siavash Mirarab. “APPLES: Scalable Distance-Based Phylogenetic Placement with or without Alignments.” Systematic Biology 69, no. 3 (2020): 566–78. https://doi.org/10.1093/sysbio/syz063
- Hasan, N.B., Biswas, A., Balaban, M., Mirarab, S., Bayzid, M.S. (2022). Fast and Accurate Branch Support Calculation for Distance-Based Phylogenetic Placements. In: Jin, L., Durand, D. (eds) Comparative Genomics. RECOMB-CG 2022. Lecture Notes in Computer Science(), vol 13234. Springer, Cham. https://doi.org/10.1007/978-3-031-06220-9_3
- Python: Version >= 3.0
Install APPLES using the following command in the command-line:
pip install apples
For listing all options, run the following command:
run_apples.py -h
Input reference (backbone) tree must be in newick format. APPLES can perform placements based on FASTA alignments of nucleotide or amino acid sequences or a distance table.
APPLES input can be either
- An alignment of query sequences to the backbone. In this case, by default, hamming distances will be computed using pairwise comparison and will be then corrected using then JC69 model. If the input is amino acid alignment, APPLES uses scoredist model instead.
- A distance matrix computed using other tools.
APPLES require a reference alignment and a query alignment. All species in the backbone tree must have a corresponding sequence in the reference alignment. You can find an example reference alignment and query alignment for ten query sequences under data/ref.fa and data/query.fa respectively.
You can run APPLES with the following command on the example input nucleotide alignment dataset:
run_apples.py -s data/ref.fa -q data/query.fa -t data/backbone.nwk
If input sequences are amino acid, use the flag -p.
The format for distance matrix is a tab delimited csv file with column and row headers. Rows should represent query sequences and columns should represent reference sequences. You can find an example distance matrix for ten query sequences under data/dist.mat. You can run APPLES on the example distance matrix by running the following command:
run_apples.py -d data/dist.mat -t data/backbone.nwk
Output is a jplace file containing placement results for all queries. For more information about jplace files, please refer to Matsen et. al. (2012) https://doi.org/10.1371/journal.pone.0031009. The output file can be specified using -o command. When output file is not specified, the result will be printed to the standard output.
Backbone tree provided to APPLES has to have its branch lengths estimated using a distance based method such as minimum evolution. This is a requirement for getting good results. We recommend FastTree2 for re-estimating branch lengths if the backbone tree is estimated using Maximum Likelihood based methods (e.g. RAxML, PASTA). We support re-estimation of branch lengths within APPLES; in fact by default, FastTree is run prior to placement using the following command.
FastTreeMP -nosupport -nt -nome -noml -log tree.log -intree backbone.nwk < ref.fa > minimum_evo_backbone.nwk
Then perform placement on the new tree:
run_apples.py -s ref.fa -q query.fa -t minimum_evo_backbone.nwk
If you already reestimated the backbone branch lengths, you can skip this step in APPLES using the option "-D" for speedup.
Please refer to the tutorial below for detailed examples of usage in alignment-based and alignment-free settings.
https://github.com/smirarab/tutorials/blob/master/Skmer-APPLES-tutorial.md
2.0.9
- Amino acid distances are now computed using the same BLOSUM45 matrix used in FastTree2.
2.0.8
- Pairwise distance overlap requirement parameter can now be changed by the user.