Purge Manual

Purge removes redundant sequences from a FASTA file. This is recommended in order to prevent highly similar sequences distorting the search for motifs. Purge was written by Andy Neuwald and is described in more detail in Neuwald et al., "Gibbs motif sampling: detection of bacterial outer membrane protein repeats", Protein Science, 4:1618--1632, 1995.

Purge works with either DNA or protein sequences and creates an output file such that no two sequences have a (gapless) local alignment score greater than a threshold specified by the user. The alignment score is based on the BLOSUM62 matrix for proteins, and on a +5/-1 scoring scheme for DNA. Purge can also be used to mask tandem repeats. It uses the XNU program for this purpose.

Using Purge

For typical-length protein sequences, a reasonable purge threshold is around 150. To purge a file of sequences named "examples/lipocalin.s", do the following command. The purged sequences will be put in file "examples/lipocalin.s.b150". You should experiment with different purge thresholds depending on your particular needs. Depending on the application values between 60 and 250 may be appropriate.
```
  purge examples/lipocalin.s 150
```
For DNA sequences of about 100 bp, a reasonable purge threshold is about 175. As the sequences become longer, higher thresholds will be required to avoid removing too many sequences. To purge a file of sequences named "examples/crp0.s", do the following command. The purged sequences will be put in file "examples/crp0.s.b175".
```
  purge examples/crp0.s 175
```

Purge Usage Message

If you type "purge" at the command line, you will receive the following usage message.

usage: purge file score 
  options:
     [-n]    - sequences are DNA (default: protein)
     [-b]    - use blast heuristic method (default for protein)
     [-e]    - use an exhaustive method (default for DNA)
     [-q]    - keep first sequence in the set
     [-x]    - use xnu to mask protein tandem repeats

  Purge creates an output file from the input file such that
  no two sequences have local alignment score greater than .
  The output file is named ..
  Substitution matrices: BLOSUM62 (protein), +5/-1 (DNA).