This script calculates the effective number of sequences (Neff) from a multiple sequence alignment (MSA) file using the following equation:

$$ Neff = \sum_{i=1}^{N} {1 \over weight_{i}} $$

where $N$ is the number of sequences in the MSA. For each sequence $i$, $weight_{i}$ is the sum of $weight_{ij}$ over all sequences in the MSA, while $weight_{ij}$ is the sequence identity between any homologous sequence $i$ and $j$ in the MSA. The minimum sequence identity between $i$ and $j$ is set to 80% by default but can be defined by the user (see below).

• Python 3
• numpy

• --msa_file: Path to an MSA file. Supported formats are .a3m, .fasta, .sto, and .hhr.
• --af_dir: Path to an AlphaFold2 output directory. This will automatically find the MSA files and calculate the total number of sequences/Neff.
• --cf_dir: Path to a ColabFold output directory. This will automatically find the MSA file and calculate the total number of sequences/Neff.
• --seq_id: Optional: Sequence identity threshold for calculating Neff. The default is 0.8 in line with the 80% value used in the AlphaFold2 paper (Jumper et al. 2021). However, be aware that other thresholds are also used: e.g. 0.62 (Wu et al. 2020).
• --input_seq: Optional: The input sequence used to calculate the MSA (fasta format). This is used to calculate the Neff/length.

python get_neff.py --msa_file <path_to_msa_file> [--seq_id <sequence_identity_threshold> --input_seq <path_to_input_sequence>]
python get_neff.py --af_dir <path_to_af_dir> [--seq_id <sequence_identity_threshold> --input_seq <path_to_input_sequence>]
python get_neff.py --cf_dir <path_to_cf_dir> [--seq_id <sequence_identity_threshold> --input_seq <path_to_input_sequence>]

Note: Can be a little slow on large MSAs (>10000 sequences)

Results are printed to the terminal in the format:

Neff = <neff value>
Neff/len = <neff/len value>
Total sequences = <total_sequences>

The results are also stored in an output file called neff.csv.

This script uses lightly modified code from AlphaFold and ColabDesign.