Algorithm for clustering protein multiple sequence alignments using normalized mutual information.
Project description
PSICalc Algorithm Package
This is a package for clustering Multiple Sequence Alignments (MSAs) utilizing normalized mutual information to examine protein subdomains. A complete data visualization tool for psicalc is available on the releases page.
As an example:
import psicalc as pc
file = "<your_fasta_file>" # e.g "PF02517_seed.txt"
data = pc.read_txt_file_format(file) # read Fasta file
data = pc.durston_schema(data, 1) # Label column index starting at 1
# If you have multiple sequences or labels, merge them first
data = pc.merge_sequences([data], ['HIST'])
result = pc.find_clusters(1, data) # will sample every column against msa
# Optionally write dictionary to csv
pc.write_output_data(1, result)
The program will run and return a csv or xlsx file with the strongest clusters found in the MSA provided.
Our initial publication can be found here: https://academic.oup.com/bioinformaticsadvances/article/2/1/vbac058/6671262
Following our initial publication, the program was found to associate invariant columns with variable columns in some cases. It was determined that the invariant columns were causing an issue, and due to their low entropy, invariant or nearly invariant positions offered little information in the way of meaningful clustering. Therefore, in the latest version (0.5.1 and beyond), we have added the ability to filter out low entropy columns using a sliding scale from 0-0.25 (0-25%) entropy where entropy is the number of different amino acids found in a column along with the number of occurrences of each amino acid. Invariant columns (i.e., those with only one amino acid) have an entropy of 0. A report of the columns removed due to low entropy is included with the output data file. As a result of these changes, data run using this latest version will not match what was found in our initial paper, but should represent clusters based upon meaningful relationships. In all cases, researchers are advised to inspect the outputs to confirm the associations are meaningful.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
