{
    "info": {
        "author": "brentp",
        "author_email": "bpederse@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "Programming Language :: Python :: 2.7",
            "Programming Language :: Python :: 3",
            "Programming Language :: Python :: 3.2",
            "Programming Language :: Python :: 3.3",
            "Topic :: Scientific/Engineering :: Bio-Informatics"
        ],
        "description": "Aclust\n======\nStreaming agglomerative clustering with custom distance and correlation\n\n\n*Agglomerative clustering* is a very simple algorithm.\nThe function `aclust` provided here is an attempt at a simple implementation\nof a modified version that allows a stream of input so that data is not\nrequired to be read into memory all at once. Most clustering algorithms operate\non a matrix of correlations which may not be feasible with high-dimensional\ndata.\n\n`aclust` **defers** some complexity to the caller by relying on a stream of\nobjects that support an interface (I know, I know) of:\n\n    obj.distance(other) -> numeric\n    obj.is_correlated(other) -> bool\n\nWhile this does add some infrastructure, we can imagine a class with\nposition and values attributes, where the former is an integer and the\nlatter is a list of numeric values. Then, those methods would be implemented\nas:\n\n    def distance(self, other):\n        return self.position - other.position\n\n    def is_correlated(self, other):\n        return np.corrcoef(self.values, other.values)[0, 1] > 0.5\n\nThis allows the `aclust` function to be used on **any** kind of data. We can\nimagine that distance might return the Levenshtein distance between 2 strings\nwhile is\\_correlated might indicate their presence in the same sentence or in\nsentences with the same sentiment.\n\nSince the input can be- and the output is- streamed, it is assumed the the objs\nare in sorted order. This is important for things like genomic data, but may be\nless so in text, where the max\\_skip parameter can be set to a large value to\ndetermine how much data is kept in memory.\n\nSee the function docstring for examples and options. The function signature is:\n\n   aclust(object\\_stream, max\\_dist,\n          max\\_skip=1, linkage='single', multi\\_member=False)\n\nIt yields clusters (lists) of objects from the input object stream.\n\n`multi\\_member` allows a feature to be a member of multiple clusters as long as\nit meets the distance and correlation constraints. The default is to only\nallow a feature to be added to the *nearest* cluster with which it is\ncorrelated.\n\nUses\n====\n\n+  Clustering methylation data which we know to be locally correlated. We can\n   use this to reduce the number of tests (of association) from 1 test per CpG,\n   to 1 test per correlated unit.\n   See: https://github.com/brentp/aclust/blob/master/examples/methylation-clustering-asthma.py for a full example.\n\n```\n    chrom   start   end n_probes   probes                asthma.pvalue   asthma.tstat    asthma.coef\n    chr1    566570  567501  8   chr1:566570,chr1:566731,chr1:567113,chr1:567206,chr1:567312,chr1:567348,chr1:567358,chr1:567501 0.4566  -0.74   -0.06\n    chr1    713985  714021  3   chr1:713985,chr1:714012,chr1:714021 0.1185  -1.56   -0.13\n    chr1    845810  846195  3   chr1:845810,chr1:846155,chr1:846195 0.5913  0.54    0.04\n    chr1    848379  848440  3   chr1:848379,chr1:848409,chr1:848440 0.3399  -0.95   -0.06\n    chr1    854766  855046  7   chr1:854766,chr1:854824,chr1:854838,chr1:854918,chr1:854951,chr1:854966,chr1:855046 0.7482  -0.32   -0.02\n    chr1    870791  871546  8   chr1:870791,chr1:870810,chr1:870958,chr1:871033,chr1:871057,chr1:871308,chr1:871441,chr1:871546 0.2198  -1.23   -0.11\n    chr1    892857  892948  3   chr1:892857,chr1:892914,chr1:892948 0.2502  -1.15   -0.05\n    chr1    901062  901799  5   chr1:901062,chr1:901449,chr1:901685,chr1:901725,chr1:901799 0.6004  0.52    0.04\n    chr1    946875  947091  4   chr1:946875,chr1:947003,chr1:947018,chr1:947091 0.9949  0.01    0.00\n```\n   So we can filter on the asthma.pvalue to find regions associated with asthma.\n  \n\nINSTALL\n=======\n\n`aclust` is available on pypi, as such it can be installed with:\n\n    pip install aclust\n\n\nAcknowledgments\n===============\n\nThe idea of this is taken from this paper:\n\n    Sofer, T., Schifano, E. D., Hoppin, J. A., Hou, L., & Baccarelli, A. A. (2013). A-clustering: A Novel Method for the Detection of Co-regulated Methylation Regions, and Regions Associated with Exposure. Bioinformatics, btt498.\n\nThe example uses a pull-request implementing GEE for python's statsmodels:\n    https://github.com/statsmodels/statsmodels/pull/928",
        "description_content_type": null,
        "docs_url": null,
        "download_url": "UNKNOWN",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/brentp/aclust/",
        "keywords": "bioinformatics cluster",
        "license": "MIT",
        "maintainer": null,
        "maintainer_email": null,
        "name": "aclust",
        "package_url": "https://pypi.org/project/aclust/",
        "platform": "any",
        "project_url": "https://pypi.org/project/aclust/",
        "project_urls": {
            "Download": "UNKNOWN",
            "Homepage": "https://github.com/brentp/aclust/"
        },
        "release_url": "https://pypi.org/project/aclust/0.1.3/",
        "requires_dist": null,
        "requires_python": null,
        "summary": "streaming agglomerative clustering",
        "version": "0.1.3"
    },
    "last_serial": 965605,
    "releases": {
        "0.1.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "4a48af6e0225b556c497d29905e13bea",
                    "sha256": "68167d9947f82b0fa9b1167271341cb05072599783091dabda450e0fa04d2ec9"
                },
                "downloads": -1,
                "filename": "aclust-0.1.0.tar.gz",
                "has_sig": false,
                "md5_digest": "4a48af6e0225b556c497d29905e13bea",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 3620,
                "upload_time": "2013-09-11T15:24:33",
                "url": "https://files.pythonhosted.org/packages/ba/f3/87e3f33a3ea3e3585025db64955a0cecb561ff2b43ce84eb4af6e2a2e4d2/aclust-0.1.0.tar.gz"
            }
        ],
        "0.1.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "d05e16de332ce8562442aa6f7f250025",
                    "sha256": "b54b3c906a461b6159481c827a6859adfb15e00ccd12f8bc7b52f57cb12bced6"
                },
                "downloads": -1,
                "filename": "aclust-0.1.2.tar.gz",
                "has_sig": false,
                "md5_digest": "d05e16de332ce8562442aa6f7f250025",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 4924,
                "upload_time": "2013-11-19T16:09:37",
                "url": "https://files.pythonhosted.org/packages/f1/25/fc2f0d64ce112753b03111948d69b4541679089a02ce784b4bea0bc82563/aclust-0.1.2.tar.gz"
            }
        ],
        "0.1.3": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "6d0041dbdc768840dbcd8e705cd85e76",
                    "sha256": "27c0d0372310ba796407cf7919001d80cb45831f31f5144d0d1f216badbc84f3"
                },
                "downloads": -1,
                "filename": "aclust-0.1.3.tar.gz",
                "has_sig": false,
                "md5_digest": "6d0041dbdc768840dbcd8e705cd85e76",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5298,
                "upload_time": "2014-01-10T17:48:32",
                "url": "https://files.pythonhosted.org/packages/02/28/5aed4aff9e1894c940348a7a1564a37cb516692d01fcbc6feb4534c0a408/aclust-0.1.3.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "6d0041dbdc768840dbcd8e705cd85e76",
                "sha256": "27c0d0372310ba796407cf7919001d80cb45831f31f5144d0d1f216badbc84f3"
            },
            "downloads": -1,
            "filename": "aclust-0.1.3.tar.gz",
            "has_sig": false,
            "md5_digest": "6d0041dbdc768840dbcd8e705cd85e76",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 5298,
            "upload_time": "2014-01-10T17:48:32",
            "url": "https://files.pythonhosted.org/packages/02/28/5aed4aff9e1894c940348a7a1564a37cb516692d01fcbc6feb4534c0a408/aclust-0.1.3.tar.gz"
        }
    ]
}