{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A Gene Ontology Tutorial in Python - Model Solutions to Exercises"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These electronic exercises (with solutions) accompany the book chapter \"A Gene Ontology Tutorial in Python\" by Alex Warwick Vesztrocy and Christophe Dessimoz, to appear in _The Gene Ontology Handbook_, C Dessimoz and N Skunca Eds, Springer Humana.\n",
    "\n",
    "Version: 1.0.2 (Feb 2019): _Updated QuickGO API calls and usage of GOATOOLS to version 0.8.12_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 2 - Querying the Gene Ontology\n",
    "\n",
    "First, we need to load the GOATools library. This enables us to parse the Gene Ontology (GO) OBO file. For more information on GOATools, see <a href=\"https://github.com/tanghaibao/goatools\">their documentation</a>."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import the OBO parser from GOATools\n",
    "from goatools import obo_parser"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to download the GO OBO file, we also require the <code>wget</code> and <code>os</code> libraries."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import wget\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we can download the OBO file into the <code>'./data'</code> folder using the following. We are going to download the <code>go-basic.obo</code> version of the ontology, which is guaranteed to be acyclic, which means that annotations can be propagated up the graph."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "go_obo_url = 'http://purl.obolibrary.org/obo/go/go-basic.obo'\n",
    "data_folder = os.getcwd() + '/data'\n",
    "\n",
    "# Check if we have the ./data directory already\n",
    "if(not os.path.isfile(data_folder)):\n",
    "    # Emulate mkdir -p (no error if folder exists)\n",
    "    try:\n",
    "        os.mkdir(data_folder)\n",
    "    except OSError as e:\n",
    "        if(e.errno != 17):\n",
    "            raise e\n",
    "else:\n",
    "    raise Exception('Data path (' + data_folder + ') exists as a file. '\n",
    "                   'Please rename, remove or change the desired location of the data path.')\n",
    "\n",
    "# Check if the file exists already\n",
    "if(not os.path.isfile(data_folder+'/go-basic.obo')):\n",
    "    go_obo = wget.download(go_obo_url, data_folder+'/go-basic.obo')\n",
    "else:\n",
    "    go_obo = data_folder+'/go-basic.obo'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The path to the GO OBO file is now stored in the variable <code>go_obo</code>."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/Users/alex/go-handbook-live/data/go-basic.obo\n"
     ]
    }
   ],
   "source": [
    "print(go_obo)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can create a dictionary of the GO terms, using the <code>obo_parser</code> from GOATools."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/Users/alex/go-handbook-live/data/go-basic.obo: fmt(1.2) rel(2019-01-27) 47,381 GO Terms\n"
     ]
    }
   ],
   "source": [
    "go = obo_parser.GODag(go_obo)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises 2.1\n",
    "#### Question 2.1.a. \n",
    "What is the name of the GO term <code>GO:00048527</code>?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.a."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO:0048527\tlevel-05\tdepth-06\tlateral root development [biological_process]\n"
     ]
    }
   ],
   "source": [
    "go_id = 'GO:0048527'\n",
    "go_term = go[go_id]\n",
    "print(go_term)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO term name: lateral root development\n",
      "GO term namespace: biological_process\n"
     ]
    }
   ],
   "source": [
    "print('GO term name: {}'.format(go_term.name))\n",
    "print('GO term namespace: {}'.format(go_term.namespace))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.1.b. \n",
    "What are the immediate parent(s) of the term <code>GO:00048527</code>?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.b."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO:0048528\tlevel-04\tdepth-05\tpost-embryonic root development [biological_process]\n"
     ]
    }
   ],
   "source": [
    "for term in go_term.parents:\n",
    "    print(term)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.1.c. \n",
    "What are the immediate children of the term <code>GO:00048527</code>?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.c."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "for term in go_term.children:\n",
    "    print(term)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.1.d. \n",
    "Recursively find all the parent and child terms of the term <code>GO:00048527</code>. _Hint_: use your solutions to the previous two questions, with a recursive loop."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.d."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we can define functions to find the parents and children of a GO term, as well as one that finds both - the transitive closure."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def transitive_closure(go_term, go):\n",
    "    go_term_set = set()\n",
    "    find_parents(go_term, go, go_term_set)\n",
    "    find_children(go_term, go, go_term_set)\n",
    "    return go_term_set\n",
    "    \n",
    "def find_parents(term1, go, go_term_set={}, ret=False):\n",
    "    for term2 in term1.parents:\n",
    "        go_term_set.update({term2})\n",
    "        \n",
    "        # Recurse on term to find all parents\n",
    "        find_parents(term2, go, go_term_set)          \n",
    "    if(ret):\n",
    "        return go_term_set\n",
    "\n",
    "def find_children(term1, go, go_term_set={}, ret=False):\n",
    "    for term2 in term1.children:\n",
    "        go_term_set.update({term2})\n",
    "        \n",
    "        # Recurse on term to find all children\n",
    "        find_children(term2, go, go_term_set)\n",
    "    if(ret):\n",
    "        return go_term_set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can create the transitive closure as a set by calling the following."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "go_term_set = transitive_closure(go_term, go)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now print this out using the \"pretty print\"f eature that is inherited from GOATools."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO:0099402\tlevel-03\tdepth-03\tplant organ development [biological_process]\n",
      "GO:0048364\tlevel-04\tdepth-04\troot development [biological_process]\n",
      "GO:0048856\tlevel-02\tdepth-02\tanatomical structure development [biological_process]\n",
      "GO:0090696\tlevel-03\tdepth-04\tpost-embryonic plant organ development [biological_process]\n",
      "GO:0032501\tlevel-01\tdepth-01\tmulticellular organismal process [biological_process]\n",
      "GO:0032502\tlevel-01\tdepth-01\tdevelopmental process [biological_process]\n",
      "GO:0048528\tlevel-04\tdepth-05\tpost-embryonic root development [biological_process]\n",
      "GO:0009791\tlevel-02\tdepth-02\tpost-embryonic development [biological_process]\n",
      "GO:0008150\tlevel-00\tdepth-00\tbiological_process [biological_process]\n"
     ]
    }
   ],
   "source": [
    "for term in go_term_set:\n",
    "    print(term)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "An alternative method is by using an inbuilt function of GOATools."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "rec = go[go_id]\n",
    "parents = rec.get_all_parents()\n",
    "children = rec.get_all_children()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO:0032502\tlevel-01\tdepth-01\tdevelopmental process [biological_process]\n",
      "GO:0048856\tlevel-02\tdepth-02\tanatomical structure development [biological_process]\n",
      "GO:0090696\tlevel-03\tdepth-04\tpost-embryonic plant organ development [biological_process]\n",
      "GO:0009791\tlevel-02\tdepth-02\tpost-embryonic development [biological_process]\n",
      "GO:0008150\tlevel-00\tdepth-00\tbiological_process [biological_process]\n",
      "GO:0099402\tlevel-03\tdepth-03\tplant organ development [biological_process]\n",
      "GO:0048528\tlevel-04\tdepth-05\tpost-embryonic root development [biological_process]\n",
      "GO:0032501\tlevel-01\tdepth-01\tmulticellular organismal process [biological_process]\n",
      "GO:0048364\tlevel-04\tdepth-04\troot development [biological_process]\n"
     ]
    }
   ],
   "source": [
    "for term in parents.union(children):\n",
    "    print(go[term])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.1.e. \n",
    "How many GO terms have the word “growth” in their name?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.e."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To do this, we can loop around every single GO term and check their name."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of GO terms with \"growth\" in their name: 626\n"
     ]
    }
   ],
   "source": [
    "growth_count = 0\n",
    "for go_term in go.values():\n",
    "    if 'growth' in go_term.name:\n",
    "        growth_count += 1\n",
    "        \n",
    "print('Number of GO terms with \"growth\" in their name: {}'.format(growth_count))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.1.f.\n",
    "What is the deepest common ancestor term of <code>GO:0048527</code> and <code>GO:0097178</code>?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.1.f."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we can write a function that finds the common GO terms."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def common_parent_go_ids(terms, go):\n",
    "    '''\n",
    "        This function finds the common ancestors in the GO \n",
    "        tree of the list of terms in the input.\n",
    "    '''\n",
    "    # Find candidates from first\n",
    "    rec = go[terms[0]]\n",
    "    candidates = rec.get_all_parents()\n",
    "    candidates.update({terms[0]})\n",
    "    \n",
    "    # Find intersection with second to nth term\n",
    "    for term in terms[1:]:\n",
    "        rec = go[term]\n",
    "        parents = rec.get_all_parents()\n",
    "        parents.update({term})\n",
    "        \n",
    "        # Find the intersection with the candidates, and update.\n",
    "        candidates.intersection_update(parents)\n",
    "        \n",
    "    return candidates"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can define the deepest common ancestor of two terms, as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def deepest_common_ancestor(terms, go):\n",
    "    '''\n",
    "        This function gets the nearest common ancestor \n",
    "        using the above function.\n",
    "        Only returns single most specific - assumes unique exists.\n",
    "    '''\n",
    "    # Take the element at maximum depth. \n",
    "    return max(common_parent_go_ids(terms, go), key=lambda t: go[t].depth)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can find the deepest common ancestor of the two terms in question. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "go_id1 = 'GO:0097178'\n",
    "go_id_id1_dca = deepest_common_ancestor([go_id, go_id1], go)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The nearest common ancestor of\n",
      "\tGO:0048527 (lateral root development)\n",
      "and\n",
      "\tGO:0097178 (ruffle assembly)\n",
      "is\n",
      "\tGO:0008150 (biological_process)\n"
     ]
    }
   ],
   "source": [
    "print('The nearest common ancestor of\\n\\t{} ({})\\nand\\n\\t{} ({})\\nis\\n\\t{} ({})'\n",
    "      .format(go_id, go[go_id].name, \n",
    "              go_id1, go[go_id1].name,\n",
    "              go_id_id1_dca, go[go_id_id1_dca].name))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises 2.2\n",
    "Using the visualisation function in GOATools, answer the following questions.\n",
    "\n",
    "#### Question 2.2.a. \n",
    "Produce a figure similar to that in Figure 1 of the chapter, for the GO term <code>GO:0097190</code>. From the visualisation, what is the name of this term?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.2.a"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We first have to create the query to return the record for the GO term <code>GO:0097190</code>."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "go_id2 = 'GO:0097192'\n",
    "rec = go[go_id2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We then need to decide where to save the image file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "lineage_png = go_id2 + '-lineage.png'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can draw the lineage, as so (you may need to install the pygraphviz library via pip if needed):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "lineage info for terms ['GO:0097192'] written to GO:0097192-lineage.png\n"
     ]
    }
   ],
   "source": [
    "go.draw_lineage([rec], lineage_img=lineage_png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Note*, the GOATools <code>draw_lineage</code> function will output a PNG to <code>./GO_lineage.png</code>, or whatever is defined by the argument <code>lineage_img</code>. This can be viewed with your system viewer, or included here by running something similar to the following code. It may be necessary to double click on the image and scroll, in order to see the detail."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(lineage_png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.2.b.\n",
    "Using this figure, what is the most specific term that is in the parent terms of both <code>GO:0097191</code> and <code>GO:0038034</code>? This is also referred to as the lowest common ancestor."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.2.b."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is possible to read off the graph - the term <code>GO:0007165</code>, _signal transduction_."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises 2.3\n",
    "Using the <code>get_term()</code> function, listed below, answer the following questions.\n",
    "\n",
    "*Note*: the <code>get_oboxml()</code> function listed in the chapter, in Source Code 2.1, will no longer work. This is due to an API overhaul of the EMBL-EBI's QuickGO browser.\n",
    "\n",
    "For the interested reader, it is also possible to use the <a href=\"https://github.com/cokelaer/bioservices\"><code>bioservices</code> library</a>, in order to retrieve information from QuickGO (as well as many other web services)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from future.standard_library import install_aliases\n",
    "install_aliases()\n",
    "from urllib.request import urlopen\n",
    "import json\n",
    "\n",
    "def get_term(go_id):\n",
    "    \"\"\"\n",
    "        This function retrieves the definition of a given Gene Ontology term,\n",
    "        using EMBL-EBI's QuickGO browser.\n",
    "        Input: go_id - a valid Gene Ontology ID, e.g. GO:0048527.\n",
    "    \"\"\"\n",
    "    quickgo_url = \"https://www.ebi.ac.uk/QuickGO/services/ontology/go/terms/\" + go_id\n",
    "    ret = urlopen(quickgo_url)\n",
    "    \n",
    "    # Check the response\n",
    "    if(ret.getcode() == 200):\n",
    "        term = json.loads(ret.read())\n",
    "        return term['results'][0]\n",
    "    else:\n",
    "        raise ValueError(\"Couldn't receive information from QuickGO. Check GO ID and try again.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.3.a. \n",
    "Find the name and description of the GO term <code>GO:0048527</code>. _Hint_: print out the dictionary returned by the function and study its structure."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.3.a.\n",
    "First, let's get the term information."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "term = get_term(go_id)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It might be useful to study the structure of this dictionary, in order to answer this question."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'id': 'GO:0048527',\n",
       " 'isObsolete': False,\n",
       " 'name': 'lateral root development',\n",
       " 'definition': {'text': 'The process whose specific outcome is the progression of the lateral root over time, from its formation to the mature structure. A lateral root is one formed from pericycle cells located on the xylem radius of the root, as opposed to the initiation of the main root from the embryo proper.'},\n",
       " 'children': [{'id': 'GO:1901332', 'relation': 'negatively_regulates'},\n",
       "  {'id': 'GO:1901333', 'relation': 'positively_regulates'},\n",
       "  {'id': 'GO:0010102', 'relation': 'part_of'},\n",
       "  {'id': 'GO:2000023', 'relation': 'regulates'},\n",
       "  {'id': 'GO:1902089', 'relation': 'part_of'}],\n",
       " 'aspect': 'biological_process',\n",
       " 'usage': 'Unrestricted'}"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "term"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can extract the relevant data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Description of GO:0048527 is:\n",
      "\tThe process whose specific outcome is the progression of the lateral root over time, from its formation to the mature structure. A lateral root is one formed from pericycle cells located on the xylem radius of the root, as opposed to the initiation of the main root from the embryo proper.\n"
     ]
    }
   ],
   "source": [
    "print('Description of {} is:\\n\\t{}'.format(go_id, term['definition']['text']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 2.3.b.\n",
    "Look at the difference in the OBO-XML output for the GO terms <code>GO:00048527</code> and <code>GO:0097178</code>, then generate a table of the synonymous relationships of the term <code>GO:0097178</code>."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 2.3.b."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Firstly, generate the OBO-XML dictionary for this new term."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "term1 = get_term(go_id1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can then extract the synonyms as follows."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'id': 'GO:0097178',\n",
       " 'isObsolete': False,\n",
       " 'name': 'ruffle assembly',\n",
       " 'definition': {'text': 'The aggregation, arrangement and bonding together of a set of components to form a ruffle, a projection at the leading edge of a crawling cell; the protrusions are supported by a microfilament meshwork. The formation of ruffles (also called membrane ruffling) is thought to be controlled by a group of enzymes known as Rho GTPases, specifically RhoA, Rac1 and cdc42.',\n",
       "  'xrefs': [{'dbCode': 'PMID', 'dbId': '12556481'},\n",
       "   {'dbCode': 'URL', 'dbId': 'http:en.wikipedia.org/wiki/Membrane_ruffling'}]},\n",
       " 'synonyms': [{'name': 'membrane ruffle formation', 'type': 'exact'},\n",
       "  {'name': 'membrane ruffling', 'type': 'exact'}],\n",
       " 'children': [{'id': 'GO:1900029', 'relation': 'positively_regulates'},\n",
       "  {'id': 'GO:1900028', 'relation': 'negatively_regulates'},\n",
       "  {'id': 'GO:1900027', 'relation': 'regulates'}],\n",
       " 'aspect': 'biological_process',\n",
       " 'usage': 'Unrestricted'}"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "term1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'name': 'membrane ruffle formation', 'type': 'exact'},\n",
       " {'name': 'membrane ruffling', 'type': 'exact'}]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "term1['synonyms']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[{'name': 'membrane ruffle formation', 'type': 'exact'}, {'name': 'membrane ruffling', 'type': 'exact'}]\n"
     ]
    }
   ],
   "source": [
    "synonyms = term1['synonyms']\n",
    "print(synonyms)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As an example, this can be nicely formatted using the <code>AsciiTable</code> module in the terminaltables package (which you may need to install)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+---------------------------+\n",
      "| Type  | Synonym                   |\n",
      "+-------+---------------------------+\n",
      "| exact | membrane ruffle formation |\n",
      "| exact | membrane ruffling         |\n",
      "+-------+---------------------------+\n"
     ]
    }
   ],
   "source": [
    "from terminaltables import AsciiTable\n",
    "\n",
    "# Initialise table data and set header\n",
    "table_data = [['Type', 'Synonym']]\n",
    "for synonym in synonyms:\n",
    "    table_data.append([synonym['type'], synonym['name']])\n",
    "print(AsciiTable(table_data).table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 3 - Retrieving GO annotations\n",
    "\n",
    "In this section we will look at how to manipulate the Gene Association File (GAF) standard, using a parser from the BioPython package."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import Bio.UniProt.GOA as GOA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we need to download a GAF file from the EBI FTP website, which hosts the current and all previous UniProt-GOA annotations. The links to these can be found on the <a href=\"https://www.ebi.ac.uk/GOA/downloads\">EBI GOA Downloads page</a>. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As an example, we are going to download the reduced GAF file containing gene association data for *Arabidopsis Thaliana*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from ftplib import FTP\n",
    "arab_uri = '/pub/databases/GO/goa/ARABIDOPSIS/goa_arabidopsis.gaf.gz'\n",
    "arab_fn = arab_uri.split('/')[-1]\n",
    "\n",
    "# Check if the file exists already\n",
    "arab_gaf = os.path.join(data_folder, arab_fn)\n",
    "if(not os.path.isfile(arab_gaf)):\n",
    "    # Login to FTP server\n",
    "    ebi_ftp = FTP('ftp.ebi.ac.uk')\n",
    "    ebi_ftp.login() # Logs in anonymously\n",
    "    \n",
    "    # Download\n",
    "    with open(arab_gaf,'wb') as arab_fp:\n",
    "        ebi_ftp.retrbinary('RETR {}'.format(arab_uri), arab_fp.write)\n",
    "        \n",
    "    # Logout from FTP server\n",
    "    ebi_ftp.quit()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can load all the annotations into a dictionary, using the iterator from the BioPython package (<code>Bio.UniProt.GOA.gafiterator</code>)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "import gzip\n",
    "\n",
    "# File is a gunzip file, so we need to open it in this way\n",
    "with gzip.open(arab_gaf, 'rt') as arab_gaf_fp:\n",
    "    arab_funcs = {}  # Initialise the dictionary of functions\n",
    "    \n",
    "    # Iterate on each function using Bio.UniProt.GOA library.\n",
    "    for entry in GOA.gafiterator(arab_gaf_fp):\n",
    "        uniprot_id = entry.pop('DB_Object_ID')\n",
    "        arab_funcs[uniprot_id] = entry"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we have a structure of the annotations which can manipulated. Each of the entries have been loaded in the following form.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'DB': 'UniProtKB', 'DB_Object_Symbol': 'PIAL1', 'Qualifier': [''], 'GO_ID': 'GO:0051176', 'DB:Reference': ['PMID:25415977'], 'Evidence': 'IGI', 'With': ['AGI_LocusCode:AT5G41580'], 'Aspect': 'P', 'DB_Object_Name': 'E4 SUMO-protein ligase PIAL1', 'Synonym': ['PIAL1', 'EMB3001', 'At1g08910', 'F7G19.21'], 'DB_Object_Type': 'protein', 'Taxon_ID': ['taxon:3702'], 'Date': '20141218', 'Assigned_By': 'TAIR', 'Annotation_Extension': '', 'Gene_Product_Form_ID': ''}\n"
     ]
    }
   ],
   "source": [
    "print(arab_funcs[list(arab_funcs.keys())[0]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises 3.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 3.1.a.\n",
    "Find the total number of annotations for Arabidopsis thaliana with <code>NOT</code> qualifiers. What is this as a percentage of the total number of annotations for this species?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 3.1.a."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is possible by looping through each of the values and checking whether <code>NOT</code> is listed as one of the qualifiers.\n",
    "\n",
    "Even though here it doesn't make a difference if we check either\n",
    "\n",
    "<code>if 'NOT' in func['Qualifier']</code>\n",
    "\n",
    "or\n",
    "\n",
    "<code>if 'NOT' in func['Qualifier'][0]</code>\n",
    "\n",
    "the first is preferred. This is because there can be multiple qualifiers for a given annotation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total count of NOT qualifiers: 802 (3.17%)\n",
      "Total number of annotations: 25289\n"
     ]
    }
   ],
   "source": [
    "not_count = 0\n",
    "total_count = len(arab_funcs)\n",
    "for func in arab_funcs.values():\n",
    "    if 'NOT' in func['Qualifier']:\n",
    "        not_count += 1\n",
    "        \n",
    "\n",
    "print('Total count of NOT qualifiers: {} ({}%)'.format(not_count, round(((float(not_count)/float(total_count))*100),2)))\n",
    "print('Total number of annotations: {}'.format(total_count))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 3.1.b.\n",
    "How many genes (of _Arabidopsis thaliana_) have the annotation <code>GO:0048527</code>?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 3.1.b."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is done by checking each annotation if its <code>GO_ID</code> entry is equal to the required term's ID.\n",
    "\n",
    "Further, here there has been a check on the taxonomic ID. This isn't strictly necessary, but would be required if the annotations database contained multiple species."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "arab_tax_id = 3702  # This isn't necessary here, but would be with the full data set.\n",
    "\n",
    "annot_count = 0\n",
    "counted_gene = []\n",
    "for uniprot_id in arab_funcs:\n",
    "    if('taxon:' + str(arab_tax_id) in arab_funcs[uniprot_id]['Taxon_ID']):\n",
    "        if((arab_funcs[uniprot_id]['GO_ID'] == go_id)):\n",
    "            counted_gene.append(uniprot_id)\n",
    "            annot_count += 1\n",
    "del counted_gene"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now find the number of genes:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17\n"
     ]
    }
   ],
   "source": [
    "print(annot_count)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 3.1.c.\n",
    "Generate a list of annotated proteins which have the word _“growth”_ in their name."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 3.1.c."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This can be generated in the following way, checking each annotation's <code>DB_Object_Name</code> field for the keyword."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "keyword = 'growth'\n",
    "growth_dict = {x: arab_funcs[x]\n",
    "               for x in arab_funcs \n",
    "               if keyword in arab_funcs[x]['DB_Object_Name']}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "UniProt IDs of annotations with \"growth\" in their name:\n",
      "\t - A0A1P8B471\n",
      "\t - B3H5J1\n",
      "\t - F4KHI3\n",
      "\t - Q3C1C7\n",
      "\t - Q3E880\n",
      "\t - Q6DSU1\n",
      "\t - Q6ID76\n",
      "\t - Q6NNL3\n",
      "\t - Q93VK8\n",
      "\t - Q9FGF6\n",
      "\t - Q9FHF5\n",
      "\t - Q9LI64\n",
      "\t - Q9SI57\n",
      "Total: 13\n"
     ]
    }
   ],
   "source": [
    "print('UniProt IDs of annotations with \"growth\" in their name:')\n",
    "for annot in growth_dict:\n",
    "    print(\"\\t - \" + annot)\n",
    "print(\"Total: {}\".format(len(growth_dict)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 3.1.d.\n",
    "There are 21 evidence codes used in the Gene Ontology project. As discussed in Chapter 3, many of these are inferred, either by curators or automatically. Find the counts of each evidence code for in the Arabidopsis thaliana annotation file."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 3.1.d."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This can be done by looping through each of the annotations, counting each of thier evidence codes and placing them into a dictionary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "evidence_count = {}\n",
    "for annotation in arab_funcs:\n",
    "    evidence = arab_funcs[annotation]['Evidence']\n",
    "    \n",
    "    if(evidence not in evidence_count):\n",
    "        evidence_count[evidence] = 1\n",
    "    else:\n",
    "        evidence_count[evidence] += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The counts are then:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-------+\n",
      "| Evidence Code | Count |\n",
      "+---------------+-------+\n",
      "| IBA           | 4730  |\n",
      "| IC            | 1     |\n",
      "| IDA           | 1911  |\n",
      "| IEA           | 11020 |\n",
      "| IEP           | 551   |\n",
      "| IGI           | 326   |\n",
      "| IMP           | 2068  |\n",
      "| IPI           | 227   |\n",
      "| ISM           | 18    |\n",
      "| ISS           | 332   |\n",
      "| NAS           | 11    |\n",
      "| ND            | 3295  |\n",
      "| RCA           | 463   |\n",
      "| TAS           | 336   |\n",
      "+---------------+-------+\n"
     ]
    }
   ],
   "source": [
    "table_data = [['Evidence Code', 'Count']]\n",
    "for code in sorted(evidence_count.keys()):\n",
    "    table_data.append([code, str(evidence_count[code])])\n",
    "print(AsciiTable(table_data).table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(All others are zero.)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 3.1.f (Extension Question).\n",
    "To help visualise the counts of each evidence code found in the previous question, produce a pie chart showing the proportion of annotations with each evidence code for the *Arabidopsis thaliana* annotations dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 3.1.f."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to draw the pie chart, we need to convert the counts into percentages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "evidence_percent = {}\n",
    "for code in evidence_count:\n",
    "        evidence_percent[code] = ((float(evidence_count[code]) /\n",
    "                                   float(total_count))\n",
    "                                  *100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can have a look at the counts and percentages for each evidence code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-------+----------------+\n",
      "| Evidence Code | Count | Percentage (%) |\n",
      "+---------------+-------+----------------+\n",
      "| IBA           | 4730  | 18.7           |\n",
      "| IC            | 1     | 0.0            |\n",
      "| IDA           | 1911  | 7.56           |\n",
      "| IEA           | 11020 | 43.58          |\n",
      "| IEP           | 551   | 2.18           |\n",
      "| IGI           | 326   | 1.29           |\n",
      "| IMP           | 2068  | 8.18           |\n",
      "| IPI           | 227   | 0.9            |\n",
      "| ISM           | 18    | 0.07           |\n",
      "| ISS           | 332   | 1.31           |\n",
      "| NAS           | 11    | 0.04           |\n",
      "| ND            | 3295  | 13.03          |\n",
      "| RCA           | 463   | 1.83           |\n",
      "| TAS           | 336   | 1.33           |\n",
      "+---------------+-------+----------------+\n"
     ]
    }
   ],
   "source": [
    "table_data = [['Evidence Code', 'Count', 'Percentage (%)']]\n",
    "for code in sorted(evidence_count.keys()):\n",
    "    table_data.append([code, str(evidence_count[code]), str(round(evidence_percent[code],2))])\n",
    "print(AsciiTable(table_data).table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we are going to plot this information as a pie chart, in order to better visualise the lack of experimental evidence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 768x768 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Declare matplotlib as inline and import pyplot\n",
    "%matplotlib inline\n",
    "from matplotlib import pyplot\n",
    "\n",
    "# Setup the figure\n",
    "fig = pyplot.figure(1, figsize=(8,8), dpi=96)\n",
    "ax=fig.add_axes([0.1,0.1,0.8,0.8])\n",
    "\n",
    "# Extract the lables / percentages\n",
    "labels = evidence_percent.keys()\n",
    "fracs = evidence_percent.values()\n",
    "\n",
    "# Make IEA obvious by \"exploding\" it\n",
    "explode = [int('IEA' in x)*0.1 for x in labels]\n",
    "\n",
    "# Plot the pie chart\n",
    "patches, texts = ax.pie(list(fracs), explode=explode, startangle=90, labeldistance=1.1) \n",
    "\n",
    "# Add \n",
    "ax.legend(patches, labels, bbox_to_anchor=(1.2, 0.75), fontsize=12)\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 4 - GO enrichment or depletion analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section, the <code>GOEnrichmentStudy()</code> function from the GOATools library will be used to perform GO enrichment analysis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "from goatools.go_enrichment import GOEnrichmentStudy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exercises 4.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Perform an enrichment analysis using the list of genes with the _\"growth\"_ keyword from exercise 3.1.c, and the GO structure from exercise 2.1."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The population is the functions observed in the _Arabidopsis thaliana_ GOA file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "pop = arab_funcs.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then, we need to create a dictionary of genes with their UniProt ID as a key and their set of GO annotations as the values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "assoc = {}\n",
    "\n",
    "for x in arab_funcs:\n",
    "    if x not in assoc:\n",
    "        assoc[x] = set()\n",
    "    assoc[x].add(str(arab_funcs[x]['GO_ID']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, the study set here is those genes with the _\"growth\"_ keyword, found previously."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "study = growth_dict.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 4.1.a.\n",
    "Which GO term is most significantly enriched or depleted? Does this make sense?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 4.1.a."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Possible methods for the GOEnrichmentStudy are:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "methods = [\"bonferroni\", \"sidak\", \"holm\", \"fdr\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is possible to run on all methods, or just a subset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fisher module not installed.  Falling back on scipy.stats.fisher_exact\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Propagating term counts to parents ..\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100% 25,289 of 25,289 population items found in association\n",
      "100%     13 of     13 study items found in association\n",
      "100%     13 of     13 study items found in population(25289)\n",
      "Calculating 5,364 uncorrected p-values using fisher_scipy_stats\n",
      "   5,364 GO terms are associated with 25,289 of 25,289 population items\n",
      "      40 GO terms are associated with     13 of     13 study items\n",
      "       3 GO terms found significant (< 0.05=alpha) after multitest correction: local bonferroni\n",
      "       3 GO terms found significant (< 0.05=alpha) after multitest correction: local sidak\n",
      "       3 GO terms found significant (< 0.05=alpha) after multitest correction: local holm\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Generate p-value distribution for FDR based on resampling (this might take a while)\n",
      "Sample 0 / 500: p-value 0.0010278710533060788\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fisher module not installed.  Falling back on scipy.stats.fisher_exact\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Sample 10 / 500: p-value 0.0015414407798976498\n",
      "Sample 20 / 500: p-value 0.0005140574953738786\n",
      "Sample 30 / 500: p-value 0.0005140574953738786\n",
      "Sample 40 / 500: p-value 3.303147547493567e-05\n",
      "Sample 50 / 500: p-value 0.0005140574953738786\n",
      "Sample 60 / 500: p-value 0.0005140574953738786\n",
      "Sample 70 / 500: p-value 0.0005140574953738786\n",
      "Sample 80 / 500: p-value 0.0076852972242090695\n",
      "Sample 90 / 500: p-value 0.0010278710533060788\n",
      "Sample 100 / 500: p-value 0.002054766781135086\n",
      "Sample 110 / 500: p-value 0.0010278710533060788\n",
      "Sample 120 / 500: p-value 0.0005140574953738786\n",
      "Sample 130 / 500: p-value 9.826609994798967e-05\n",
      "Sample 140 / 500: p-value 0.000318804720807801\n",
      "Sample 150 / 500: p-value 0.0010278710533060788\n",
      "Sample 160 / 500: p-value 4.150813978478498e-05\n",
      "Sample 170 / 500: p-value 0.0010278710533060788\n",
      "Sample 180 / 500: p-value 0.00026027744906719874\n",
      "Sample 190 / 500: p-value 0.003593283493731873\n",
      "Sample 200 / 500: p-value 0.0010278710533060788\n",
      "Sample 210 / 500: p-value 0.002054766781135086\n",
      "Sample 220 / 500: p-value 0.0005140574953738786\n",
      "Sample 230 / 500: p-value 0.0010278710533060788\n",
      "Sample 240 / 500: p-value 0.0015414407798976498\n",
      "Sample 250 / 500: p-value 0.004105635653855762\n",
      "Sample 260 / 500: p-value 0.002054766781135086\n",
      "Sample 270 / 500: p-value 0.002054766781135086\n",
      "Sample 280 / 500: p-value 0.0005140574953738786\n",
      "Sample 290 / 500: p-value 0.0005140574953738786\n",
      "Sample 300 / 500: p-value 0.003593283493731873\n",
      "Sample 310 / 500: p-value 0.0005140574953738786\n",
      "Sample 320 / 500: p-value 0.000160821606967919\n",
      "Sample 330 / 500: p-value 0.0005140574953738786\n",
      "Sample 340 / 500: p-value 0.0005140574953738786\n",
      "Sample 350 / 500: p-value 0.0005140574953738786\n",
      "Sample 360 / 500: p-value 3.303147547493567e-05\n",
      "Sample 370 / 500: p-value 0.0005140574953738786\n",
      "Sample 380 / 500: p-value 0.0010278710533060788\n",
      "Sample 390 / 500: p-value 0.0010278710533060788\n",
      "Sample 400 / 500: p-value 0.0010278710533060788\n",
      "Sample 410 / 500: p-value 0.0010278710533060788\n",
      "Sample 420 / 500: p-value 0.0005140574953738786\n",
      "Sample 430 / 500: p-value 0.0010278710533060788\n",
      "Sample 440 / 500: p-value 0.0005140574953738786\n",
      "Sample 450 / 500: p-value 0.0005140574953738786\n",
      "Sample 460 / 500: p-value 0.0010278710533060788\n",
      "Sample 470 / 500: p-value 0.0005140574953738786\n",
      "Sample 480 / 500: p-value 0.0010278710533060788\n",
      "Sample 490 / 500: p-value 0.0010278710533060788\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       5 GO terms found significant (< 0.05=alpha) after multitest correction: local fdr\n"
     ]
    }
   ],
   "source": [
    "g = GOEnrichmentStudy(pop, assoc, go,\n",
    "                         propagate_counts=True,\n",
    "                         alpha=0.05,\n",
    "                         methods=methods)\n",
    "g_res = g.run_study(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GO\tNS\tenrichment\tname\tratio_in_study\tratio_in_pop\tp_uncorrected\tdepth\tstudy_count\tp_bonferroni\tp_sidak\tp_holm\tp_fdr\tstudy_items\n",
      "GO:0030154\tBP\te\tcell differentiation          \t6/13\t57/25289\t1.69e-13\t3\t6\t9.09e-10\t8.86e-10\t9.09e-10\t0\tQ3E880, Q6DSU1, Q6ID76, Q6NNL3, Q9LI64, Q9SI57\n",
      "GO:0048869\tBP\te\tcellular developmental process\t6/13\t113/25289\t1.16e-11\t2\t6\t6.24e-08\t6.09e-08\t6.24e-08\t0\tQ3E880, Q6DSU1, Q6ID76, Q6NNL3, Q9LI64, Q9SI57\n",
      "GO:0032502\tBP\te\tdevelopmental process         \t7/13\t712/25289\t2.02e-08\t1\t7\t0.000108\t0.000105\t0.000108\t0\tB3H5J1, Q3E880, Q6DSU1, Q6ID76, Q6NNL3, Q9LI64, Q9SI57\n",
      "GO:2000012\tBP\te\tregulation of auxin polar transport\t2/13\t20/25289\t4.61e-05\t5\t2\t0.247\t0.241\t0.247\t0.014\t Q93VK8, Q9FGF6\n",
      "GO:0051049\tBP\te\tregulation of transport       \t2/13\t42/25289\t0.000208\t4\t2\t1\t1\t1\t0.036\t Q93VK8, Q9FGF6\n",
      "GO:0010817\tBP\te\tregulation of hormone levels  \t2/13\t69/25289\t0.000561\t3\t2\t1\t1\t1\t0.422\t Q93VK8, Q9FGF6\n",
      "GO:0032879\tBP\te\tregulation of localization    \t2/13\t107/25289\t0.00134\t3\t2\t1\t1\t1\t0.638\t Q93VK8, Q9FGF6\n",
      "GO:0065008\tBP\te\tregulation of biological quality\t2/13\t265/25289\t0.00791\t2\t2\t1\t1\t1\t0.988\t Q93VK8, Q9FGF6\n",
      "GO:0048527\tBP\te\tlateral root development      \t1/13\t17/25289\t0.00871\t6\t1\t1\t1\t1\t0.99\t         B3H5J1\n"
     ]
    }
   ],
   "source": [
    "g.print_results(g_res, min_ratio=None, pval=0.01)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the most significant term is <code>GO:0030154</code> (cell differentiation). It makes sense that this would be associated with proteins having the keyword _\"growth\"_ in their name."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 4.1.b.\n",
    "How many terms are enriched, when using the Bonferroni corrected method and a p-value $\\le$ 0.01?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 4.1.b."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Perform the GO Enrichment study using the Bonferroni corrected method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fisher module not installed.  Falling back on scipy.stats.fisher_exact\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Propagating term counts to parents ..\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100% 25,289 of 25,289 population items found in association\n",
      "100%     13 of     13 study items found in association\n",
      "100%     13 of     13 study items found in population(25289)\n",
      "Calculating 5,364 uncorrected p-values using fisher_scipy_stats\n",
      "   5,364 GO terms are associated with 25,289 of 25,289 population items\n",
      "      40 GO terms are associated with     13 of     13 study items\n",
      "       3 GO terms found significant (< 0.01=alpha) after multitest correction: local bonferroni\n"
     ]
    }
   ],
   "source": [
    "g_bonferroni = GOEnrichmentStudy(pop, assoc, go,\n",
    "                                 propagate_counts=True,\n",
    "                                 alpha=0.01,\n",
    "                                 methods=['bonferroni'])\n",
    "g_bonferroni_res = g_bonferroni.run_study(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "s_bonferroni = []\n",
    "for x in g_bonferroni_res:\n",
    "    if x.p_bonferroni <= 0.01:\n",
    "        s_bonferroni.append((x.goterm.id, x.p_bonferroni))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('GO:0030154', 9.088903406049323e-10), ('GO:0048869', 6.244383841758757e-08), ('GO:0032502', 0.00010815519774707948)]\n"
     ]
    }
   ],
   "source": [
    "print(s_bonferroni)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 4.1.c.\n",
    "How many terms are enriched with false discovery rate (a.k.a. q-value) $\\le$ 0.01?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 4.1.c."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fisher module not installed.  Falling back on scipy.stats.fisher_exact\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Propagating term counts to parents ..\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100% 25,289 of 25,289 population items found in association\n",
      "100%     13 of     13 study items found in association\n",
      "100%     13 of     13 study items found in population(25289)\n",
      "Calculating 5,364 uncorrected p-values using fisher_scipy_stats\n",
      "   5,364 GO terms are associated with 25,289 of 25,289 population items\n",
      "      40 GO terms are associated with     13 of     13 study items\n",
      "fisher module not installed.  Falling back on scipy.stats.fisher_exact\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Generate p-value distribution for FDR based on resampling (this might take a while)\n",
      "Sample 0 / 500: p-value 0.011253078458717746\n",
      "Sample 10 / 500: p-value 0.0005140574953738786\n",
      "Sample 20 / 500: p-value 0.0010278710533060788\n",
      "Sample 30 / 500: p-value 0.0005140574953738786\n",
      "Sample 40 / 500: p-value 0.0005140574953738786\n",
      "Sample 50 / 500: p-value 0.0008701371457487843\n",
      "Sample 60 / 500: p-value 0.0005140574953738786\n",
      "Sample 70 / 500: p-value 0.0005140574953738786\n",
      "Sample 80 / 500: p-value 0.0015414407798976498\n",
      "Sample 90 / 500: p-value 0.0010278710533060788\n",
      "Sample 100 / 500: p-value 0.0005140574953738786\n",
      "Sample 110 / 500: p-value 0.003593283493731873\n",
      "Sample 120 / 500: p-value 0.0010278710533060788\n",
      "Sample 130 / 500: p-value 0.0005140574953738786\n",
      "Sample 140 / 500: p-value 0.006663750638263605\n",
      "Sample 150 / 500: p-value 0.003080688032113681\n",
      "Sample 160 / 500: p-value 0.0015414407798976498\n",
      "Sample 170 / 500: p-value 0.0010278710533060788\n",
      "Sample 180 / 500: p-value 0.004105635653855762\n",
      "Sample 190 / 500: p-value 0.0010278710533060788\n",
      "Sample 200 / 500: p-value 0.0005140574953738786\n",
      "Sample 210 / 500: p-value 0.0010278710533060788\n",
      "Sample 220 / 500: p-value 0.003593283493731873\n",
      "Sample 230 / 500: p-value 0.0005140574953738786\n",
      "Sample 240 / 500: p-value 0.0010278710533060788\n",
      "Sample 250 / 500: p-value 0.0005140574953738786\n",
      "Sample 260 / 500: p-value 0.0005140574953738786\n",
      "Sample 270 / 500: p-value 0.0010278710533060788\n",
      "Sample 280 / 500: p-value 0.003080688032113681\n",
      "Sample 290 / 500: p-value 0.0005140574953738786\n",
      "Sample 300 / 500: p-value 0.0031108913301437103\n",
      "Sample 310 / 500: p-value 0.0010278710533060788\n",
      "Sample 320 / 500: p-value 0.0010278710533060788\n",
      "Sample 330 / 500: p-value 0.0005140574953738786\n",
      "Sample 340 / 500: p-value 0.0010278710533060788\n",
      "Sample 350 / 500: p-value 0.011761793778736767\n",
      "Sample 360 / 500: p-value 0.0010278710533060788\n",
      "Sample 370 / 500: p-value 0.0005140574953738786\n",
      "Sample 380 / 500: p-value 0.0005140574953738786\n",
      "Sample 390 / 500: p-value 0.0005140574953738786\n",
      "Sample 400 / 500: p-value 0.011253078458717746\n",
      "Sample 410 / 500: p-value 0.0005140574953738786\n",
      "Sample 420 / 500: p-value 0.0005140574953738786\n",
      "Sample 430 / 500: p-value 0.0010278710533060788\n",
      "Sample 440 / 500: p-value 0.0005140574953738786\n",
      "Sample 450 / 500: p-value 0.0010278710533060788\n",
      "Sample 460 / 500: p-value 0.0010278710533060788\n",
      "Sample 470 / 500: p-value 0.0005140574953738786\n",
      "Sample 480 / 500: p-value 0.002567849163328763\n",
      "Sample 490 / 500: p-value 0.0005140574953738786\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       5 GO terms found significant (< 0.05=alpha) after multitest correction: local fdr\n"
     ]
    }
   ],
   "source": [
    "g_fdr = GOEnrichmentStudy(pop, assoc, go,\n",
    "                          propagate_counts=True,\n",
    "                          alpha=0.05,\n",
    "                          methods=['fdr'])\n",
    "g_fdr_res = g_fdr.run_study(study)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "s_fdr = []\n",
    "for x in g_fdr_res:\n",
    "    if x.p_fdr <= 0.01:\n",
    "        s_fdr.append((x.goterm.id, x.p_fdr))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('GO:0030154', 0.0), ('GO:0048869', 0.0), ('GO:0032502', 0.0), ('GO:2000012', 0.006)]\n"
     ]
    }
   ],
   "source": [
    "print(s_fdr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 5 - Computing basic semantic similarities between GO terms"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section we look at how to compute semantic similarity between GO terms."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "class TermCounts():\n",
    "    '''\n",
    "        TermCounts counts the term counts for each \n",
    "    '''\n",
    "    def __init__(self, go, annots):\n",
    "        '''\n",
    "            Initialise the counts and  \n",
    "        '''\n",
    "        # Backup\n",
    "        self._go = go\n",
    "        \n",
    "        # Initialise the counters\n",
    "        self._counts = Counter()\n",
    "        self._aspect_counts = Counter()\n",
    "        \n",
    "        # Fill the counters...\n",
    "        self._count_terms(go, annots)\n",
    "        \n",
    "    def _count_terms(self, go, annots):\n",
    "        '''\n",
    "            Fills in the counts and overall aspect counts.\n",
    "        '''\n",
    "        for x in annots:\n",
    "            # Extract term information\n",
    "            go_id = annots[x]['GO_ID']\n",
    "            namespace = go[go_id].namespace\n",
    "\n",
    "            self._counts[go_id] += 1\n",
    "            rec = go[go_id]\n",
    "            parents = rec.get_all_parents()\n",
    "            for p in parents:\n",
    "                self._counts[p] += 1\n",
    "            \n",
    "            self._aspect_counts[namespace] += 1\n",
    "            \n",
    "    def get_count(self, go_id):\n",
    "        '''\n",
    "            Returns the count of that GO term observed in the annotations.\n",
    "        '''\n",
    "        return self._counts[go_id]\n",
    "        \n",
    "    def get_total_count(self, aspect):\n",
    "        '''\n",
    "            Gets the total count that's been precomputed.\n",
    "        '''\n",
    "        return self._aspect_counts[aspect]\n",
    "    \n",
    "    def get_term_freq(self, go_id):\n",
    "        '''\n",
    "            Returns the frequency at which a particular GO term has \n",
    "            been observed in the annotations.\n",
    "        '''\n",
    "        try:\n",
    "            namespace = self._go[go_id].namespace\n",
    "            freq = float(self.get_count(go_id)) / float(self.get_total_count(namespace))\n",
    "        except ZeroDivisionError:\n",
    "            freq = 0\n",
    "        \n",
    "        return freq"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 5.1.a.\n",
    "<code>GO:0048364</code> (root development) and <code>GO:0044707</code> (single-multicellular organism process) are two GO terms taken from Figure 1. Calculate the semantic similarity between them based on the inverse of the semantic distance (number of branches separating them)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 5.1.a."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we need to write a function that calculates the minimum number of branches connecting two GO terms."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "def min_branch_length(go_id1, go_id2, go):\n",
    "    '''\n",
    "        Finds the minimum branch length between two terms in the GO DAG.\n",
    "    '''\n",
    "    # First get the deepest common ancestor\n",
    "    dca = deepest_common_ancestor([go_id1, go_id2], go)\n",
    "    \n",
    "    # Then get the distance from the DCA to each term\n",
    "    dca_depth = go[dca].depth\n",
    "    d1 = go[go_id1].depth - dca_depth\n",
    "    d2 = go[go_id2].depth - dca_depth\n",
    "    \n",
    "    # Return the total distance - i.e., to the deepest common ancestor and back.\n",
    "    return d1 + d2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "go_id3 = 'GO:0048364'\n",
    "go_id4 = 'GO:0044707'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can calculate the semantic distance and semantic similarity, as so:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "def semantic_distance(go_id1, go_id2, go):\n",
    "    '''\n",
    "        Finds the semantic distance (minimum number of connecting branches) \n",
    "        between two GO terms.\n",
    "    '''\n",
    "    return min_branch_length(go_id1, go_id2, go)\n",
    "\n",
    "def semantic_similarity(go_id1, go_id2, go):\n",
    "    '''\n",
    "        Finds the semantic similarity (inverse of the semantic distance) \n",
    "        between two GO terms.\n",
    "    '''\n",
    "    return 1.0 / float(semantic_distance(go_id1, go_id2, go))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The semantic similarity between terms GO:0048364 and GO:0044707 is 0.2.\n"
     ]
    }
   ],
   "source": [
    "sim = semantic_similarity(go_id3, go_id4, go)\n",
    "print('The semantic similarity between terms {} and {} is {}.'.format(go_id3, go_id4, sim))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 5.1.b.\n",
    "Calculate the  the information content (IC) of the GO term <code>GO:0048364</code> (root development), based on the frequency of observation in _Arabidopsis thaliana_."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 5.1.b."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we need to define what the information content is."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "\n",
    "def ic(go_id, termcounts):\n",
    "    '''\n",
    "        Calculates the information content of a GO term.\n",
    "    '''\n",
    "    # Get the observed frequency of the GO term\n",
    "    freq = termcounts.get_term_freq(go_id)\n",
    "\n",
    "    # Calculate the information content (i.e., -log(\"freq of GO term\")\n",
    "    return -1.0 * math.log(freq)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can calculate the information content of the single term, <code>GO:0048364</code>."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "# First get the counts of each GO term.\n",
    "termcounts = TermCounts(go, arab_funcs)\n",
    "\n",
    "# Calculate the information content\n",
    "infocontent = ic(go_id, termcounts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.703332455042499\n"
     ]
    }
   ],
   "source": [
    "print(infocontent)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Question 5.1.c.\n",
    "Calculate the Resnik similarity measure between the same two terms as in 5.1.a."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Answer 5.1.c."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Recall that Resnik's similarity measure is defined as the information content of the most informative common ancestor. That is, the most specific common parent-term in the GO."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def resnik_sim(go_id1, go_id2, go, termcounts):\n",
    "    '''\n",
    "        Computes Resnik's similarity measure.\n",
    "    '''\n",
    "    msca = deepest_common_ancestor([go_id1, go_id2], go)\n",
    "    return ic(msca, termcounts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can calculate this as follows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "sim_r = resnik_sim(go_id3, go_id4, go, termcounts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Resnik similarity score: -0.0\n"
     ]
    }
   ],
   "source": [
    "print('Resnik similarity score: {}'.format(sim_r))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}