{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Similarity scores and representative covers\n",
    "\n",
    "Compare ligands and protein interfaces, explore similarity scores, and find representative clusters. {ref}`Similarity scores <similarity-score-reference>` defines the metrics and files."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "from plinder.core import PlinderRelease, query_table\n",
    "from plinder.core.scores import (\n",
    "    reconstruct_interface_similarity_scores,\n",
    "    reconstruct_similarity_scores,\n",
    ")\n",
    "\n",
    "release = PlinderRelease()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ligand_system_id = \"4agi__1__1.C__1.R\"\n",
    "ligand_id = \"4agi__1__1.R\"\n",
    "interface_id = \"7cm8__1__1.A--2.A\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compare ligand similarities\n",
    "\n",
    "`reconstruct_similarity_scores()` reads the files needed for the requested query and target IDs. This example uses a one-pair self-comparison to keep execution small."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ligand_scores = reconstruct_similarity_scores(\n",
    "    query_system_ids=[ligand_system_id],\n",
    "    target_system_ids=[ligand_system_id],\n",
    "    query_ligand_ids=[ligand_id],\n",
    "    target_ligand_ids=[ligand_id],\n",
    ")\n",
    "ligand_scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For larger neighborhoods, read selected columns and rows from the complete ligand score table."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_ligand_neighborhood(system_id, ligand_id):\n",
    "    return pd.read_parquet(\n",
    "        release.fetch(\"ligand_similarity_scores\"),\n",
    "        columns=[\n",
    "            \"query_system\",\n",
    "            \"query_ligand_id\",\n",
    "            \"target_system\",\n",
    "            \"target_ligand_id\",\n",
    "            \"pocket_qcov\",\n",
    "            \"pocket_fident_qcov\",\n",
    "            \"pli_qcov\",\n",
    "            \"sucos_shape\",\n",
    "        ],\n",
    "        filters=[\n",
    "            (\"query_system\", \"==\", system_id),\n",
    "            (\"query_ligand_id\", \"==\", ligand_id),\n",
    "        ],\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compare protein-interface similarities\n",
    "\n",
    "The interface API likewise accepts selected query and target IDs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "interface_scores = reconstruct_interface_similarity_scores(\n",
    "    query_interface_ids=[interface_id],\n",
    "    target_interface_ids=[interface_id],\n",
    ")\n",
    "interface_scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Representative covers\n",
    "\n",
    "Request cover columns through `query_table()` just like annotation columns. {ref}`Representative covers <representative-cover-reference>` describes their construction and long-form files."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pocket_cover = \"pocket_qcov__70__ligand__directed_set_cover\"\n",
    "ligand_cover = query_table(\n",
    "    columns=[\n",
    "        \"system_id\",\n",
    "        \"ligand_id\",\n",
    "        pocket_cover,\n",
    "        f\"{pocket_cover}__is_centroid\",\n",
    "    ],\n",
    "    filters=[(\"ligand_id\", \"==\", ligand_id)],\n",
    ")\n",
    "ligand_cover.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "interface_cover = query_table(\n",
    "    \"interface_annotations\",\n",
    "    columns=[\n",
    "        \"system_id\",\n",
    "        \"interface_qcov__70__directed_set_cover\",\n",
    "    ],\n",
    "    filters=[(\"system_id\", \"==\", interface_id)],\n",
    ")\n",
    "interface_cover.T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Matched molecular pairs\n",
    "\n",
    "Use `ligand_mmp_pairs` to look up transformations involving the selected ligand chemistry."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "smiles_id = query_table(\n",
    "    columns=[\"ligand_smiles_id\"],\n",
    "    filters=[(\"ligand_id\", \"==\", ligand_id)],\n",
    ").iloc[0, 0]\n",
    "mmp_neighbors = query_table(\n",
    "    \"ligand_mmp_pairs\",\n",
    "    filters=[[\n",
    "        (\"ligand_smiles_id_1\", \"==\", smiles_id),\n",
    "        (\"ligand_smiles_id_2\", \"==\", smiles_id),\n",
    "    ]],\n",
    ")\n",
    "mmp_neighbors.head()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}