/psimpl

Peptide-Spectrum match IMPutation Library

Primary LanguagePythonApache License 2.0Apache-2.0

Peptide-Spectrum match IMPutation Library (PSIMPL)

PSIMPL is a python library for easy imputation of missing PSM data. Given an input PIN file, PSIMPL determines which features contain missing values. PSIMPL then performs imputation by training a regressor on fully observed feature samples.

PSIMPL's API offers a high degree of flexibility, including a large number of state-of-the-art regression algorithms such as: -Linear Regression -Ridge Regression -Lasso (i.e., L1 regularization) -ElasticNet (i.e., L1+L2 regularization)

Future work: cross-validation (CV) for hyperparameter optimization, XGBoost/DNN/Support Vector Regression