This project is read-only.



Epipred optionalParameters inputFile outputFile
Epipred -help

For each line of input, reports the most likely epitope(s). In case of ties, combines result into one line.

where the optional parameters are
  • -inputHasHeader
  • -merLength k
  • -withinDOfCenter d
  • -hlaSet [singleton OR supertype OR all]
  • -modelOnly
  • -showBy [all OR hla OR length OR hlaAndLength OR doNotGroup]
  • -model LANLIEDB03062007

Web Version

The web version of this tool uses a different predictive model.

This desktop version offers more options for scanning.


Each line of the input should be of the form

peptide <TAB> hla ... [when "-hlaSet singleton" is given]
peptide <TAB> supertype ...[when "-hlaSet supertype" is given]
peptide ... [when "-hlaSet all" is given]

Additional fields will be echoed to output. If a header is given, it
must have columns for at least peptide and, if given, hla or supertype

Sample Input

For the LANLIEDB03062007 model, HLA's must be of the form of A,B, or C followed by two digits, except for HLAs that start A68 or B15 which must be of four digits.

Supertype table used by the LANLIEDB03062007 model

Optional Parameters

  • If "-inputHasHeader" is given, the first line of the input is threated as a header.
  • "-merLength k", where k is "scan" (default), "8", "9", "10", "11", or "given"
    • With "scan" all lengths are scanned.
    • With a numeric value, only mers of that length are scanned
    • With "given" there is no scanning; the whole input peptide is scored.
    • In all cases, an adjustment is made to correct for the relative frequency of the length/hla pair.
  • "-withinDOfCenter d" where d is an integer or null (default)
    • When an integer this limits the scan for epitopes to mers such that
at least one edge of the mer is within d amino acids of the center
of the peptide given for scanning. If the input peptide has even
length, we use left of center for center.
  • -hlaSet
    • "-hlaSet singleton" (default) means to use the hla from the file
    • "-hlaSet supertype" means use the supertype from the file, scanning all its hlas
    • "-hlaSet all" means to scan all known hlas
Among the hlas considered, the best will be reported
  • "-modelOnly" if given then uses the model's probability even if a peptide is on a source list.
Otherwise, peptides on a source list are given a probability of 1.0.
(The "source list" is the list of epitopes and HLAs used to train the model.
The source list for model LANLIEDB03062007
  • "-showBy" tells how to report the maximum probability prediction(s)
    • "-showBy all" (default) - for each input line, reports the maximum probability prediction(s).
    • "-showBy hla" - for each input line and hla, reports.
    • "-showBy length" - for each input line and peptide length, reports.
    • "-showBy hlaAndLength" - for each input line and hla and peptide length, reports.
    • "-showBy doNotGroup" - for each input line, reports every prediction.


The output is a header line followed by one output line for each input line.
Additional fields are inserted after the required inputs, the fields are:
  • HLA (if hla is not given)
  • PosteriorProbability
  • WeightOfEvidence
  • BestNFlank (for the LANLIEDB03062007 model, always empty because it doesn't use flanking regions)
  • BestEpitope
  • BestCFlank (for the LANLIEDB03062007 model, always empty because it doesn't use flanking regions)
  • EpitopeLength
  • FlankingLength (for the LANLIEDB03062007 model, always 0)
  • EpitiopeStartPosition
  • EpitopeLastPosition
  • Source (for the LANLIEDB03062007 model, LANL and/or IEDB)

Sample Output


D. Heckerman, C. Kadie, and J. Listgarten, "Leveraging Information Across HLA Alleles/Supertypes Improves HLA-Specific Epitope Prediction," RECOMB 2006.

Last edited Apr 19, 2007 at 11:31 PM by CarlK, version 15


No comments yet.