epistasis-v2 exceptions reference¶

All exceptions raised by epistasis-v2 inherit from a single base class, EpistasisError. You can catch the base class to handle any library error, or catch a specific subclass to handle a particular failure mode.

from epistasis.exceptions import EpistasisError, FittingError, XMatrixError

Exception hierarchy¶

Exception
└── EpistasisError
    ├── FittingError
    └── XMatrixError

EpistasisError¶

The base class for all epistasis-v2 errors. Inherits directly from the built-in Exception.

Catch EpistasisError when you want a single handler for any problem the library raises, regardless of its specific cause.

from epistasis.exceptions import EpistasisError

try:
    model.fit()
    predictions = model.predict()
except EpistasisError as exc:
    print(f"epistasis error: {exc}")

FittingError¶

Raised when a model fit has failed or when model methods that require a completed fit are called before fit() has been invoked.

Common triggers:

• Calling predict() or score() before fit()
• Reading model.epistasis.thetas before fit()
• A nonlinear minimizer failing to converge (raises FittingError rather than surfacing the raw lmfit error)

from epistasis.exceptions import FittingError

try:
    predictions = model.predict()
except FittingError:
    # Model has not been fitted yet, fit it first
    model.fit()
    predictions = model.predict()

XMatrixError¶

Raised for an invalid or missing epistasis design matrix. This typically means a model property that depends on the design matrix was accessed before add_gpm() was called.

Common triggers:

• Accessing model.gpm, model.epistasis, or model.Xcolumns before calling model.add_gpm(gpm)
• Passing a design matrix with incompatible shape or type

from epistasis.exceptions import XMatrixError

try:
    cols = model.Xcolumns
except XMatrixError:
    # GPM not yet attached, wire it up
    model.add_gpm(gpm)
    cols = model.Xcolumns

Handling both errors¶

In practice, you may encounter either error in sequence: XMatrixError if the GPM is missing, then FittingError if the model has not been fitted.

from epistasis.exceptions import FittingError, XMatrixError

try:
    predictions = model.predict()
except XMatrixError:
    model.add_gpm(gpm)
    model.fit()
    predictions = model.predict()
except FittingError:
    model.fit()
    predictions = model.predict()