Pytorch refactor

examples/refactor/ALR.ipynb

@@ -0,0 +1,378 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "461ff352",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%load_ext autoreload"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "e536cc44",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%autoreload 1"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "acadee5d",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<torch._C.Generator at 0x7ff5805ca0f0>"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "import biom\n",
+    "import torch\n",
+    "import torch.nn as nn\n",
+    "%aimport mmvec.ALR\n",
+    "\n",
+    "import numpy as np\n",
+    "\n",
+    "torch.manual_seed(15)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "14d8bfab",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "microbes = biom.load_table(\"./soil_microbes.biom\")\n",
+    "metabolites = biom.load_table(\"./soil_metabolites.biom\")\n",
+    "\n",
+    "model = mmvec.ALR.MMvecALR(microbes, metabolites, 15, sigma_u=1, sigma_v=1, holdout_num=5)\n",
+    "\n",
+    "microbes = microbes.to_dataframe().T\n",
+    "metabolites = metabolites.to_dataframe().T\n",
+    "microbes = microbes.loc[metabolites.index]\n",
+    "\n",
+    "microbe_idx = microbes.columns\n",
+    "metabolite_idx = metabolites.columns\n",
+    "microbes = torch.tensor(microbes.values, dtype=torch.int)\n",
+    "metabolites = torch.tensor(metabolites.values, dtype=torch.int64)\n",
+    "\n",
+    "microbe_relative_frequency = (microbes.T/microbes.sum(1)).T\n",
+    "\n",
+    "microbe_count = microbes.shape[1]\n",
+    "\n",
+    "metabolite_count = metabolites.shape[1]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "9003e38d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#model = mmvec.ALR.MMvecALR(microbe_count, metabolite_count, 15, sigma_u=1, sigma_v=1)\n",
+    "learning_rate = 1e-3\n",
+    "batch_size = 500\n",
+    "epochs = 3000\n",
+    "optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, maximize=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "29799ea5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.microbe_relative_freq(model.microbes_train)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "b977e212",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "maybe = mmvec.train.mmvec_training_loop(\n",
+    "                            model=model,\n",
+    "\n",
+    "                            batch_size=batch_size,\n",
+    "                            epochs=epochs,\n",
+    "                            learning_rate=learning_rate,\n",
+    "    summary_interval = 25)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "30db58c0",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<generator object mmvec_training_loop at 0x7ff58099d900>"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "test_stats = pd.Dataframe.from_records(maybe, )"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2eb531fe",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1f27301f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.microbe_relative_freq(model.microbes_train)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "8a16a24a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "l"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3663ecb4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# h = model.ranks_df - model.ranks_df.mean(axis=0)\n",
+    "h = model.ranks_matrix\n",
+    "\n",
+    "k = model.latent_dim"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6b3bcb21",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "h.mean(dim=0).shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "52e80094",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from torch.linalg import svd\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b24646d3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "u, s, v = svd(h, full_matrices=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "afc2f855",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "u.shape, s.shape, v.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5b427a72",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "s"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "752acb7a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.U"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "18824f01",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.get_ordination()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a5f97b6d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ranks = pd.DataFrame(model.ranks(),\n",
+    "                     index=microbe_idx,\n",
+    "                     columns=metabolite_idx)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "99a455e5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ranks"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "fe702e51",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for param in model.parameters():\n",
+    "    print(param.shape)\n",
+    "\n",
+    "print(model.parameters)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "76a6f81b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "a = torch.randint(10, (9, 3), dtype=torch.float)\n",
+    "b = torch.randint(10, (3,))\n",
+    "a, a - a.mean(dim=0), a.mean(dim=0)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "aa9e8f31",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bp = model.get_ordination()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6b850001",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os.path"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "8e3b2153",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bp_path = os.path.join(\"/Users/keeganevans/Desktop/\", \"biplot\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b18f859b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bp.write(bp_path)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "34569491",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.ranks_df"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "46d440f7",
+   "metadata": {},
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.8.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}