Tolerance Identification API

The Tolerance Identification API finds the top N closest (by Blosum62) N-mers in the human genome against a given protein of sequence.

Examples

Get the top 10 closest 9-mers:

lev engine submit tolerance-identification NLYIQWLKDGGPSSGRPPPS --top-n 10

Get the top 5 closet 9 and 15-mers:

lev engine submit tolerance-identification NLYIQWLKDGGPSSGRPPPS --top-n 5 --nmer-sizes 9,15

Inputs

• --sequence (str)
  • Input protein of sequence to compare against

Options

• --top-n (int)
  • Collect the top N matches
  • default = 20
• --nmer-sizes
  • Nmer size(s) to run this on (Comma separated string ex: 9,10,11,12)
  • default = 9

Outputs

• out.csv
  • CSV file containing the following columns
    • nmer_size - size of this nmer
    • resnum - residue number (1 indexed) of the nmer position in the query sequence
    • query_seq - query sequence
    • matchrank - Rank (0=best, N = worst ) out of the top-N closest (by blosum62) nmers to the query
    • matchscore - blosum62 score of the result to the query sequence
    • matchseq - the found human genome sequence
    • matchscore/max_score - matchscore divided by the score of a 100% (normalized Blosum62)
• out.json
  • JSON format of the out.csv

Notes

Running this protocol takes between 4 and 5 GB of memory per CPU

Input proteome

The input proteome file was taken from https://ftp.ensembl.org/pub/release-110/fasta/homo_sapiens/pep/Homo_sapiens.GRCh38.pep.abinitio.fa.gz