{
    "info": {
        "author": "",
        "author_email": "",
        "bugtrack_url": null,
        "classifiers": [],
        "description": "# Fr\u00e9chet ChemNet Distance\n\nThe new wave of successful generative models in machine learning has increased\nthe interest in deep learning driven de novo drug design. However, assessing\nthe performance of such generative models is notoriously difficult. Metrics that\nare typically used to assess the performance of such generative models are the\npercentage of chemically valid molecules or the similarity to real molecules in\nterms of particular descriptors, such as the partition coefficient (logP) or druglike-\nness. However, method comparison is difficult because of the inconsistent use of\nevaluation metrics, the necessity for multiple metrics, and the fact that some of\nthese measures can easily be tricked by simple rule-based systems. We propose a\nnovel distance measure between two sets of molecules, called Fr\u00e9chet ChemNet\ndistance (FCD), that can be used as an evaluation metric for generative models. The\nFCD is similar to a recently established performance metric for comparing image\ngeneration methods, the Fr\u00e9chet Inception Distance (FID). Whereas the FID uses\none of the hidden layers of InceptionNet, the FCD utilizes the penultimate layer\nof a deep neural network called \u201cChemNet\u201d, which was trained to predict drug\nactivities. Thus, the FCD metric takes into account chemically and biologically\nrelevant information about molecules, and also measures the diversity of the set\nvia the distribution of generated molecules. The FCD\u2019s advantage over previous\nmetrics is that it can detect if generated molecules are a) diverse and have similar\nb) chemical and c) biological properties as real molecules. We further provide an\neasy-to-use implementation that only requires the SMILES representation of the\ngenerated molecules as input to calculate the FCD.\n\n\n\n",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/bioinf-jku/FCD",
        "keywords": "",
        "license": "LGPLv3",
        "maintainer": "",
        "maintainer_email": "",
        "name": "FCD",
        "package_url": "https://pypi.org/project/FCD/",
        "platform": "",
        "project_url": "https://pypi.org/project/FCD/",
        "project_urls": {
            "Homepage": "https://github.com/bioinf-jku/FCD"
        },
        "release_url": "https://pypi.org/project/FCD/1.0/",
        "requires_dist": [
            "keras",
            "numpy",
            "scipy",
            "tensorflow",
            "rdkit ; extra == 'rdkit'"
        ],
        "requires_python": "",
        "summary": "Fr\u00e9chet ChEMNet Distance",
        "version": "1.0"
    },
    "last_serial": 4546305,
    "releases": {
        "1.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "5933590ae865985b7821c287f2a7e3e5",
                    "sha256": "1379efcd7610e64f19f3744ef1a9f9cfa39a98e0c6b4d619f466735ecc2e8600"
                },
                "downloads": -1,
                "filename": "FCD-1.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "5933590ae865985b7821c287f2a7e3e5",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 53120859,
                "upload_time": "2018-11-30T08:57:27",
                "url": "https://files.pythonhosted.org/packages/2e/bb/5d1e60f30c74b17918e2b62853d0fcba32c5de70c218a5e52358acaa3f55/FCD-1.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "482490fb34e41f5fa4771fb71d178297",
                    "sha256": "7fb300ed1190ae7666fd859136f37444e7f5da2db939b68475a2647d75db19da"
                },
                "downloads": -1,
                "filename": "FCD-1.0.tar.gz",
                "has_sig": false,
                "md5_digest": "482490fb34e41f5fa4771fb71d178297",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 53118192,
                "upload_time": "2018-11-30T08:57:39",
                "url": "https://files.pythonhosted.org/packages/3c/4f/0e667259805dc68cc6aeba178f3f5938ab3b974de8309e286b21589a557f/FCD-1.0.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "5933590ae865985b7821c287f2a7e3e5",
                "sha256": "1379efcd7610e64f19f3744ef1a9f9cfa39a98e0c6b4d619f466735ecc2e8600"
            },
            "downloads": -1,
            "filename": "FCD-1.0-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "5933590ae865985b7821c287f2a7e3e5",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 53120859,
            "upload_time": "2018-11-30T08:57:27",
            "url": "https://files.pythonhosted.org/packages/2e/bb/5d1e60f30c74b17918e2b62853d0fcba32c5de70c218a5e52358acaa3f55/FCD-1.0-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "482490fb34e41f5fa4771fb71d178297",
                "sha256": "7fb300ed1190ae7666fd859136f37444e7f5da2db939b68475a2647d75db19da"
            },
            "downloads": -1,
            "filename": "FCD-1.0.tar.gz",
            "has_sig": false,
            "md5_digest": "482490fb34e41f5fa4771fb71d178297",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 53118192,
            "upload_time": "2018-11-30T08:57:39",
            "url": "https://files.pythonhosted.org/packages/3c/4f/0e667259805dc68cc6aeba178f3f5938ab3b974de8309e286b21589a557f/FCD-1.0.tar.gz"
        }
    ]
}