Skip to content

Licensed applications

Some InterProScan analyses depend on third-party applications that must be licensed and installed separately.

You do not need a separate license to run InterProScan itself or to analyse your own sequences with the bundled analyses.

In InterProScan 6, the licensed applications are:

  • DeepTMHMM
  • Phobius
  • SignalP-Euk
  • SignalP-Prok

These analyses remain unavailable until you install the third-party software and configure InterProScan to use the extracted installation directory.

Tip

You only need to download and configure the licensed applications you plan to run.

Activation steps

To enable a licensed application in InterProScan:

  1. Obtain a license from the software provider.
  2. Download and extract the application archive.
  3. Create a Nextflow config file, for example licensed.conf.
  4. Add the absolute path to the extracted directory under params.appsConfig.
  5. Run InterProScan with -c licensed.conf.

Obtaining the software

DeepTMHMM 1.0

Request a standalone copy of DeepTMHMM 1.0 by emailing licensing@biolib.com.

After receiving the archive:

unzip -q DeepTMHMM-v1.0.zip

Phobius 1.01

Download Phobius 1.01 from Erik Sonnhammer's website, then extract it:

tar -zxf phobius101_linux.tgz

Info

Phobius skips sequences containing the non-standard or ambiguous residues O, B, Z, or J. Other InterProScan analyses continue to run normally for those sequences.

SignalP 6.0

SignalP 6.0 is available in two model variants:

  • Full (slow)
  • Distilled (fast)

Use the full model when you need the most accurate region boundaries and have sufficient time and memory available.

Licenses and downloads are available from the DTU SignalP 6.0 page.

After downloading the archive:

tar -zxf signalp-6.0i.fast.tar.gz

Configuration

Store the application paths in a config file such as licensed.conf:

params {
  appsConfig {
    deeptmhmm {
      dir = "/full/path/to/DeepTMHMM"
    }
    phobius {
      dir = "/full/path/to/phobius"
    }
    signalp_euk {
      dir = "/full/path/to/signal6p_fast"
      mode = "fast"
    }
    signalp_prok {
      dir = "/full/path/to/signal6p_fast"
      mode = "fast"
    }
  }
}

Danger

For SignalP, set mode = "fast" for the distilled model, and mode = "slow" or mode = "slow-sequential" for the full model. slow runs the full model in parallel and requires substantially more memory. slow-sequential runs the same full model sequentially, uses roughly the same memory as fast, and is about six times slower.

Run InterProScan with the default analyses plus all licensed deep-learning analyses:

nextflow run ebi-pf-team/interproscan6 \
  -r 6.0.1 \
  -profile docker \
  -c licensed.conf \
  --input /path/to/sequences.faa \
  --run-ml  # (1)!
  1. DeepTMHMM, SignalP-Euk, and SignalP-Prok use deep-learning models that are computationally intensive. They must be enabled explicitly with --run-ml, or selected explicitly with --applications.

Run only the licensed applications explicitly:

nextflow run ebi-pf-team/interproscan6 \
  -r 6.0.1 \
  -profile docker \
  -c licensed.conf \
  --input /path/to/sequences.faa \
  --applications deeptmhmm,phobius,signalp_euk,signalp_prok

Info

Running both signalp_euk and signalp_prok executes SignalP twice, once with the eukaryotic model and once with the prokaryotic model. When possible, run only the model that matches your dataset.