simulated_msa v1.0: simulated multiple sequence alignments with known phylogenies for benchmarking MSA tools

This dataset contains 1,860 simulated multiple sequence alignments (MSAs) with known phylogenies. The alignments were generated using the AliSim tool from IQ-TREE v2.4.0.

The simulation parameters span a broad range of conditions:

• Number of sequences: 1,000–100,000
• Substitution models: LG, JTT, and WAG
• Sequence lengths: 400–2,000 residues
• Sequence identities: 8%–75%
• Gap fractions: 0%–99%

Directory structure

The dataset contains three main directories:

• fasta/ – unaligned protein sequences [FASTA format]
• msa/ – aligned protein sequences (reference MSAs) [FASTA format]
• tree/ – phylogenetic trees corresponding to each simulated MSA (reference trees) [Newick format]

A metadata file (metadata.tsv) is also included, providing detailed information for each simulated MSA

Metadata

A metadata file (metadata.tsv) is included, containing detailed information for each simulated MSA. It provides:

• id – unique MSA identifier
• seqs_count – number of sequences in the MSA
• alisim_length – seed sequence length
• alisim_rlen_min / mean / max – relative branch length parameters
• alisim_ins / alisim_del – insertion and deletion rates
• alisim_model – substitution model (LG, JTT, WAG)
• alisim_model_type – model configuration used by AliSim
• mean_identity_percent – average sequence identity [%]
• mean_gaps_percent – average fraction of gaps [%]
• min_seq_length / mean_seq_length / max_seq_length – sequence length statistics