{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import wot"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "# Notebook 2: Computing transport matrices\n",
    "\n",
    "In this notebook we compute transport matrices connecting each pair of time-points, and we examine the effect of each parameter on the solution. \n",
    "\n",
    "To compute the transport matrix $\\pi_{t_1,t_2}$ connecting cells $x_1, \\ldots, x_n$ at time $t_1$ to cells $y_1, \\ldots, y_m$ at time $t_2$, we solve an optimization problem over all matrices $\\pi$ that obey certain row-sum and column-sum constraints. \n",
    "These constraints ensure that the total amount of mass flowing out of each cell $x_i$ and into each cell $y_j$ adds up the correct amount. \n",
    "We select the transport matrix with the lowest possible transport cost, subject to these constraints. \n",
    "\n",
    "The amount of mass flowing out of each cell $x_i$ depends on the *growth rate* of the cell. We do this in step 1 below, before proceeding to computing transport maps in step 2. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "# Step 1: Construct initial estimate of cell growth rates (optional)\n",
    "\n",
    "Before computing transport maps, we first form an initial estimate of the cell growth rate function $g(x)$. \n",
    "We do this using signatures of proliferation and apoptosis computed in Notebook 1. \n",
    "Note that we refine this initial estimate using unbalanced optimal transport, as explained later in this notebook. \n",
    "\n",
    "We model cellular growth with a Birth-Death Process, which assigns each cell $x$ a *rate* of division $\\beta(x)$ and a  rate of death $\\delta(x)$. \n",
    "In a small interval of time $dt$, these *rates* imply that the probability of a division is $\\beta(x)dt$, and the probability of death is $\\delta(x) dt$. \n",
    "The expected number of descendants after time $dt = t_2 - t_1$ is $e^{dt(\\beta(x) - \\delta(x))}$.\n",
    "\n",
    "To compute the birth (i.e. division) rate and death rate, we apply a logistic function to transform the gene set scores into biologically plausible vaules for our reprogramming dataset (where the most proliferative cells double every 9 hours, and the least proliferative cells double every 60 hours). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "# load proliferation and apoptosis scores\n",
    "gene_set_scores = pd.read_csv('data/gene_set_scores.csv', index_col=0)\n",
    "proliferation=gene_set_scores['Cell.cycle']\n",
    "apoptosis = gene_set_scores['Apoptosis']\n",
    "\n",
    "# apply logistic function to transform to birth rate and death rate\n",
    "def logistic(x, L, k, x0=0):\n",
    "    f = L / (1 + np.exp(-k * (x - x0)))\n",
    "    return f\n",
    "def gen_logistic(p, beta_max, beta_min, pmax, pmin, center, width):\n",
    "    return beta_min + logistic(p, L=beta_max - beta_min, k=4 / width, x0=center)\n",
    "\n",
    "def beta(p, beta_max=1.7, beta_min=0.3, pmax=1.0, pmin=-0.5, center=0.25):\n",
    "    return gen_logistic(p, beta_max, beta_min, pmax, pmin, center, width=0.5)\n",
    "\n",
    "def delta(a, delta_max=1.7, delta_min=0.3, amax=0.5, amin=-0.4, center=0.1):\n",
    "    return gen_logistic(a, delta_max, delta_min, amax, amin, center,\n",
    "                          width=0.2)\n",
    "\n",
    "birth = beta(proliferation)\n",
    "death = delta(apoptosis)\n",
    "\n",
    "# growth rate is given by \n",
    "gr = np.exp(birth-death)\n",
    "growth_rates_df = pd.DataFrame(index=gene_set_scores.index, data={'cell_growth_rate':gr})\n",
    "growth_rates_df.to_csv('data/growth_gs_init.txt')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "j1JdUgTXawEJ",
    "pycharm": {}
   },
   "source": [
    "# Step 2: Compute transport maps\n",
    "\n",
    "We are now ready to compute transport maps. \n",
    "\n",
    "The next code block reads in the expression matrix, cell days, and the initial cell growth rates we computed in step 1 above. If the cell growth rates are not provided, the algorithm will assume each cell grows at roughly the same rate.  Note that the expression matrix we use here contains variable gene expression data of 1,400 genes from two time-course experiments, and we apply a filter to select cells from the serum time course. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "OgHGRa2aIax6",
    "pycharm": {}
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(175472, 1479)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "VAR_GENE_DS_PATH = 'data/ExprMatrix.var.genes.h5ad'\n",
    "CELL_DAYS_PATH = 'data/cell_days.txt'\n",
    "SERUM_CELL_IDS_PATH = 'data/serum_cell_ids.txt'\n",
    "CELL_GROWTH_PATH = 'data/growth_gs_init.txt'\n",
    "\n",
    "# load data\n",
    "adata = wot.io.read_dataset(VAR_GENE_DS_PATH, obs=[CELL_DAYS_PATH, CELL_GROWTH_PATH], obs_filter=SERUM_CELL_IDS_PATH)\n",
    "adata.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eXxmqmQua1I0",
    "pycharm": {}
   },
   "source": [
    "# The optimization problem\n",
    "\n",
    "We solve the following *unbalanced transport* optimization problem introduced in Chizat et al 2018, where we only enforce the row-sum constraints approximately and we add entropy to the transport matrix: \n",
    "\n",
    "\\begin{equation}\n",
    "\\begin{aligned}\n",
    "\\underset{\\pi}{\\text{minimize}} & \\qquad \\iint c(x,y) \\pi(x,y) dx dy - \\epsilon \\int \\pi(x,y) \\log \\pi(x,y) dx dy \\\\\n",
    "&\\qquad + \\lambda_2 {\\text{KL}} \\left ( \\int \\pi(x,y) dx  \\Big \\vert d \\hat {\\mathbb{P}}_{t_2} (y) \\int g(x)^{t_2 - t_1} d \\hat {\\mathbb{P}}_{t_1}(x) \\right ) \\\\ \n",
    "& \\qquad + \\lambda_1 {\\text{KL}} \\left ( \\int \\pi(x,y) dy \\Big \\vert d \\hat {\\mathbb{P}}_{t_1} (x) g(x)^{t_2 - t_1} \\right). \n",
    "\\end{aligned}\n",
    "\\end{equation}\n",
    "\n",
    "Here we use the notation  $\\hat {\\mathbb{P}}_{t_k} = \\frac 1 n \\sum_{i=1}^n \\delta_{x_i}$ for the empirical distribution of samples $x_1,\\ldots,x_n$ at time $t_k$, and $\\text{KL}(P \\vert Q)$ denotes the KL-divergence between distributions $P$ and $Q$. The function $c(x,y)$ encodes the cost of transporting a unit mass from $x$ to $y$. We define $c(x,y)$ to be the squared euclidean distance between cells in local PCA space. This PCA space is computed separately for each pair of time-points. Finally, the function $g(x)$ encodes the growth rate of cell $x$, and is used to specify the budget of descendant mass for each cell $x_i$ at time $t_1$. \n",
    "\n",
    "The optimization problem has three regularization parameters: \n",
    "* $\\epsilon$ controls the degree of entropy in the transport map. \n",
    "A larger value gives more entropic descendant distributions, where cells are able to obtain more fates. \n",
    "* $\\lambda_1$ controls the constraint on the row sums of $\\pi_{t_1,t_2}$, which depend on the growth rate function $g(x)$ \n",
    "A smaller value of $\\lambda_1$ enforces the constraints less strictly, which is useful when we do not have precise information about $g(x)$.\n",
    "* $\\lambda_2$ controls the constraint on the column sums of $\\pi_{t_1,t_2}$.\n",
    "\n",
    "In the following code block we initialize an OTModel, which is parameterized by $\\epsilon, \\lambda_1, \\lambda_2$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1558,
     "status": "ok",
     "timestamp": 1556313217324,
     "user": {
      "displayName": "Joshua Gould",
      "photoUrl": "",
      "userId": "09978885733763067346"
     },
     "user_tz": 240
    },
    "id": "YKuGivoIca9G",
    "outputId": "c0639c77-d5ed-4e9e-9130-b994bfad5e9c",
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "# create OTModel\n",
    "ot_model = wot.ot.OTModel(adata,epsilon = 0.05, lambda1 = 1,lambda2 = 50) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now use the OTModel to compute transport maps between any pair of time-points. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compute a single transport map from day 7 to 7.5\n",
    "tmap_annotated = ot_model.compute_transport_map(7,7.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "The object tmap_annotated is an annotated transport matrix, and tmap_annotated.X contains the actual numerical matrix. \n",
    "Both rows and columns are annotated with cell barcodes and cell growth rates (just along rows). These annotations are accessed as follows: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "pycharm": {},
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>g0</th>\n",
       "      <th>g1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>index</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGAGGAATCGC-1</th>\n",
       "      <td>1.268895</td>\n",
       "      <td>0.977589</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGAGTAGATGT-1</th>\n",
       "      <td>2.978660</td>\n",
       "      <td>2.510199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCAACCGCCA-1</th>\n",
       "      <td>1.871918</td>\n",
       "      <td>1.789104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCACTATCTT-1</th>\n",
       "      <td>3.072989</td>\n",
       "      <td>2.790364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCAGCCTGTG-1</th>\n",
       "      <td>1.026225</td>\n",
       "      <td>0.918584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCATATACGC-1</th>\n",
       "      <td>1.412582</td>\n",
       "      <td>1.366065</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCATCCGTGG-1</th>\n",
       "      <td>1.182596</td>\n",
       "      <td>1.144585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGTCGGAATCT-1</th>\n",
       "      <td>0.982687</td>\n",
       "      <td>1.047557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGGTAGTACCT-1</th>\n",
       "      <td>0.417277</td>\n",
       "      <td>0.443068</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTCTTATG-1</th>\n",
       "      <td>0.348761</td>\n",
       "      <td>0.282756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTGGGCCA-1</th>\n",
       "      <td>2.636122</td>\n",
       "      <td>2.038604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTGTCAAG-1</th>\n",
       "      <td>0.399207</td>\n",
       "      <td>0.446110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGAGAGCTC-1</th>\n",
       "      <td>1.564738</td>\n",
       "      <td>1.329046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGAGTACAT-1</th>\n",
       "      <td>1.485496</td>\n",
       "      <td>1.384023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGATCCGAG-1</th>\n",
       "      <td>0.699411</td>\n",
       "      <td>0.789952</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGGGCACTA-1</th>\n",
       "      <td>1.776569</td>\n",
       "      <td>1.709204</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGCACTTAACG-1</th>\n",
       "      <td>0.990188</td>\n",
       "      <td>0.850556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGGTCCCTTGT-1</th>\n",
       "      <td>3.447298</td>\n",
       "      <td>3.750008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCCGCAAGC-1</th>\n",
       "      <td>0.405894</td>\n",
       "      <td>0.475874</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCGCGTTTC-1</th>\n",
       "      <td>1.291190</td>\n",
       "      <td>1.144758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCGTGGACC-1</th>\n",
       "      <td>3.467186</td>\n",
       "      <td>2.937608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGCAAAGAGTACCG-1</th>\n",
       "      <td>2.457092</td>\n",
       "      <td>1.622737</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGCAATCACTTATC-1</th>\n",
       "      <td>1.255358</td>\n",
       "      <td>1.328561</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGAGTTGGC-1</th>\n",
       "      <td>1.755248</td>\n",
       "      <td>1.891669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGCTAACTC-1</th>\n",
       "      <td>1.652005</td>\n",
       "      <td>1.428341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGGAGTACC-1</th>\n",
       "      <td>0.543827</td>\n",
       "      <td>0.574121</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGCAGTTTACG-1</th>\n",
       "      <td>2.881975</td>\n",
       "      <td>2.747883</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGCATTAACCG-1</th>\n",
       "      <td>1.117299</td>\n",
       "      <td>0.860455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGGTACTCGCG-1</th>\n",
       "      <td>2.765129</td>\n",
       "      <td>2.744694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGTCTCGATGA-1</th>\n",
       "      <td>1.426713</td>\n",
       "      <td>1.493590</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCAAACTGCT-1</th>\n",
       "      <td>0.546127</td>\n",
       "      <td>0.668312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCACATGACT-1</th>\n",
       "      <td>3.406330</td>\n",
       "      <td>3.133376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCATACTCTT-1</th>\n",
       "      <td>2.149502</td>\n",
       "      <td>1.922778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCGTGGCCCTA-1</th>\n",
       "      <td>2.583077</td>\n",
       "      <td>2.744961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCGTTTAAGCC-1</th>\n",
       "      <td>3.267880</td>\n",
       "      <td>2.653814</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCAATAAGG-1</th>\n",
       "      <td>0.681796</td>\n",
       "      <td>0.808452</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCATTATCC-1</th>\n",
       "      <td>3.685506</td>\n",
       "      <td>2.870278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCCTTGACC-1</th>\n",
       "      <td>0.337586</td>\n",
       "      <td>0.440907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCCTTTCTC-1</th>\n",
       "      <td>2.856774</td>\n",
       "      <td>2.429800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCTTGGGTA-1</th>\n",
       "      <td>0.425422</td>\n",
       "      <td>0.620381</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCCATCGATTG-1</th>\n",
       "      <td>0.267748</td>\n",
       "      <td>0.358732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCGTAGTACCT-1</th>\n",
       "      <td>2.577453</td>\n",
       "      <td>2.360288</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCGTTATGCGT-1</th>\n",
       "      <td>0.959450</td>\n",
       "      <td>1.076300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCCTTAATC-1</th>\n",
       "      <td>3.275354</td>\n",
       "      <td>3.232358</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCGATGAGG-1</th>\n",
       "      <td>0.470076</td>\n",
       "      <td>0.660028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCTGTTTGT-1</th>\n",
       "      <td>2.642529</td>\n",
       "      <td>2.428539</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGCGCTCTTACCGC-1</th>\n",
       "      <td>0.257352</td>\n",
       "      <td>0.273753</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTAGCGAGAAA-1</th>\n",
       "      <td>0.294904</td>\n",
       "      <td>0.404630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTAGGACACCA-1</th>\n",
       "      <td>0.512239</td>\n",
       "      <td>0.583090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTCAAAGGTGC-1</th>\n",
       "      <td>1.495061</td>\n",
       "      <td>1.318807</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTGTGATGATA-1</th>\n",
       "      <td>2.325406</td>\n",
       "      <td>2.188960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTGTTGAACTC-1</th>\n",
       "      <td>0.321137</td>\n",
       "      <td>0.383881</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTTCGTACCGG-1</th>\n",
       "      <td>1.524528</td>\n",
       "      <td>1.443170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTTCTAACGGT-1</th>\n",
       "      <td>0.792568</td>\n",
       "      <td>0.755125</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAAGTGTCCCG-1</th>\n",
       "      <td>0.841414</td>\n",
       "      <td>0.897749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAAGTTAGCGG-1</th>\n",
       "      <td>0.338531</td>\n",
       "      <td>0.370486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCACAGTGGAGT-1</th>\n",
       "      <td>0.582584</td>\n",
       "      <td>0.609164</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAGTAATTGGA-1</th>\n",
       "      <td>0.254926</td>\n",
       "      <td>0.296702</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCATCGTGACAT-1</th>\n",
       "      <td>0.594715</td>\n",
       "      <td>0.876372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCATCTCCCTGA-1</th>\n",
       "      <td>1.198471</td>\n",
       "      <td>1.539571</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6507 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    g0        g1\n",
       "index                                           \n",
       "D7_Dox_C1_AAACCTGAGGAATCGC-1  1.268895  0.977589\n",
       "D7_Dox_C1_AAACCTGAGTAGATGT-1  2.978660  2.510199\n",
       "D7_Dox_C1_AAACCTGCAACCGCCA-1  1.871918  1.789104\n",
       "D7_Dox_C1_AAACCTGCACTATCTT-1  3.072989  2.790364\n",
       "D7_Dox_C1_AAACCTGCAGCCTGTG-1  1.026225  0.918584\n",
       "D7_Dox_C1_AAACCTGCATATACGC-1  1.412582  1.366065\n",
       "D7_Dox_C1_AAACCTGCATCCGTGG-1  1.182596  1.144585\n",
       "D7_Dox_C1_AAACCTGTCGGAATCT-1  0.982687  1.047557\n",
       "D7_Dox_C1_AAACGGGGTAGTACCT-1  0.417277  0.443068\n",
       "D7_Dox_C1_AAACGGGTCTCTTATG-1  0.348761  0.282756\n",
       "D7_Dox_C1_AAACGGGTCTGGGCCA-1  2.636122  2.038604\n",
       "D7_Dox_C1_AAACGGGTCTGTCAAG-1  0.399207  0.446110\n",
       "D7_Dox_C1_AAAGATGAGAGAGCTC-1  1.564738  1.329046\n",
       "D7_Dox_C1_AAAGATGAGAGTACAT-1  1.485496  1.384023\n",
       "D7_Dox_C1_AAAGATGAGATCCGAG-1  0.699411  0.789952\n",
       "D7_Dox_C1_AAAGATGAGGGCACTA-1  1.776569  1.709204\n",
       "D7_Dox_C1_AAAGATGCACTTAACG-1  0.990188  0.850556\n",
       "D7_Dox_C1_AAAGATGGTCCCTTGT-1  3.447298  3.750008\n",
       "D7_Dox_C1_AAAGATGTCCGCAAGC-1  0.405894  0.475874\n",
       "D7_Dox_C1_AAAGATGTCGCGTTTC-1  1.291190  1.144758\n",
       "D7_Dox_C1_AAAGATGTCGTGGACC-1  3.467186  2.937608\n",
       "D7_Dox_C1_AAAGCAAAGAGTACCG-1  2.457092  1.622737\n",
       "D7_Dox_C1_AAAGCAATCACTTATC-1  1.255358  1.328561\n",
       "D7_Dox_C1_AAAGTAGAGAGTTGGC-1  1.755248  1.891669\n",
       "D7_Dox_C1_AAAGTAGAGCTAACTC-1  1.652005  1.428341\n",
       "D7_Dox_C1_AAAGTAGAGGAGTACC-1  0.543827  0.574121\n",
       "D7_Dox_C1_AAAGTAGCAGTTTACG-1  2.881975  2.747883\n",
       "D7_Dox_C1_AAAGTAGCATTAACCG-1  1.117299  0.860455\n",
       "D7_Dox_C1_AAAGTAGGTACTCGCG-1  2.765129  2.744694\n",
       "D7_Dox_C1_AAAGTAGTCTCGATGA-1  1.426713  1.493590\n",
       "...                                ...       ...\n",
       "D7_Dox_C2_TTTATGCCAAACTGCT-1  0.546127  0.668312\n",
       "D7_Dox_C2_TTTATGCCACATGACT-1  3.406330  3.133376\n",
       "D7_Dox_C2_TTTATGCCATACTCTT-1  2.149502  1.922778\n",
       "D7_Dox_C2_TTTATGCGTGGCCCTA-1  2.583077  2.744961\n",
       "D7_Dox_C2_TTTATGCGTTTAAGCC-1  3.267880  2.653814\n",
       "D7_Dox_C2_TTTATGCTCAATAAGG-1  0.681796  0.808452\n",
       "D7_Dox_C2_TTTATGCTCATTATCC-1  3.685506  2.870278\n",
       "D7_Dox_C2_TTTATGCTCCTTGACC-1  0.337586  0.440907\n",
       "D7_Dox_C2_TTTATGCTCCTTTCTC-1  2.856774  2.429800\n",
       "D7_Dox_C2_TTTATGCTCTTGGGTA-1  0.425422  0.620381\n",
       "D7_Dox_C2_TTTCCTCCATCGATTG-1  0.267748  0.358732\n",
       "D7_Dox_C2_TTTCCTCGTAGTACCT-1  2.577453  2.360288\n",
       "D7_Dox_C2_TTTCCTCGTTATGCGT-1  0.959450  1.076300\n",
       "D7_Dox_C2_TTTCCTCTCCTTAATC-1  3.275354  3.232358\n",
       "D7_Dox_C2_TTTCCTCTCGATGAGG-1  0.470076  0.660028\n",
       "D7_Dox_C2_TTTCCTCTCTGTTTGT-1  2.642529  2.428539\n",
       "D7_Dox_C2_TTTGCGCTCTTACCGC-1  0.257352  0.273753\n",
       "D7_Dox_C2_TTTGGTTAGCGAGAAA-1  0.294904  0.404630\n",
       "D7_Dox_C2_TTTGGTTAGGACACCA-1  0.512239  0.583090\n",
       "D7_Dox_C2_TTTGGTTCAAAGGTGC-1  1.495061  1.318807\n",
       "D7_Dox_C2_TTTGGTTGTGATGATA-1  2.325406  2.188960\n",
       "D7_Dox_C2_TTTGGTTGTTGAACTC-1  0.321137  0.383881\n",
       "D7_Dox_C2_TTTGGTTTCGTACCGG-1  1.524528  1.443170\n",
       "D7_Dox_C2_TTTGGTTTCTAACGGT-1  0.792568  0.755125\n",
       "D7_Dox_C2_TTTGTCAAGTGTCCCG-1  0.841414  0.897749\n",
       "D7_Dox_C2_TTTGTCAAGTTAGCGG-1  0.338531  0.370486\n",
       "D7_Dox_C2_TTTGTCACAGTGGAGT-1  0.582584  0.609164\n",
       "D7_Dox_C2_TTTGTCAGTAATTGGA-1  0.254926  0.296702\n",
       "D7_Dox_C2_TTTGTCATCGTGACAT-1  0.594715  0.876372\n",
       "D7_Dox_C2_TTTGTCATCTCCCTGA-1  1.198471  1.539571\n",
       "\n",
       "[6507 rows x 2 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# row annotations include cell growth rates\n",
    "tmap_annotated.obs\n",
    "\n",
    "# columns annotated by cell barcodes\n",
    "# tmap_annotated.var"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The column `g0` displays the input growth rates and the column `g1` displays the row sums of the transport map (raised to the power $t_2 - t_1$. These values can differ, as we see in the plot below, because the row-sum constraints are only enforced loosely. We interpret the value `g1` as an *inferred growth rate*. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize how growth rates change with growth iterations\n",
    "plt.scatter(tmap_annotated.obs['g0'],tmap_annotated.obs['g1'])\n",
    "plt.xlabel(\"g0\")\n",
    "plt.ylabel(\"g1\")\n",
    "plt.title(\"Input vs Output Growth Rates\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {}
   },
   "source": [
    "When we increase the regularization parameter $\\lambda_1$, the row sum constraints are enforced more strictly. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "ot_model_strict = wot.ot.OTModel(adata,epsilon = 0.05, lambda1 = 3,lambda2 = 50) \n",
    "tmap_anno_strict = ot_model_strict.compute_transport_map(7,7.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(tmap_anno_strict.obs['g0'],tmap_anno_strict.obs['g1'])\n",
    "plt.xlabel(\"g0\")\n",
    "plt.ylabel(\"g1\")\n",
    "plt.title(\"Input vs Output Growth Rates\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Updating growth estimates\n",
    "\n",
    "We can feed these new growth rates back into the OT solver and recompute transport maps with these new growth rates. Instead of doing this manually, we can specify the number of 'growth iterations' with the flag `growth_iters`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "ot_model_gr2 = wot.ot.OTModel(adata,epsilon = 0.05, lambda1 = 1,lambda2 = 50,growth_iters=2) \n",
    "tmap_anno_gr2 = ot_model_gr2.compute_transport_map(7,7.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The row annotations of this transport map now have fields for multiple iterations of learning growth:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>g0</th>\n",
       "      <th>g1</th>\n",
       "      <th>g2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>index</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGAGGAATCGC-1</th>\n",
       "      <td>1.268895</td>\n",
       "      <td>0.977589</td>\n",
       "      <td>0.771923</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGAGTAGATGT-1</th>\n",
       "      <td>2.978660</td>\n",
       "      <td>2.510199</td>\n",
       "      <td>2.130938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCAACCGCCA-1</th>\n",
       "      <td>1.871918</td>\n",
       "      <td>1.789104</td>\n",
       "      <td>1.692330</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCACTATCTT-1</th>\n",
       "      <td>3.072989</td>\n",
       "      <td>2.790364</td>\n",
       "      <td>2.516863</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCAGCCTGTG-1</th>\n",
       "      <td>1.026225</td>\n",
       "      <td>0.918584</td>\n",
       "      <td>0.815139</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCATATACGC-1</th>\n",
       "      <td>1.412582</td>\n",
       "      <td>1.366065</td>\n",
       "      <td>1.303655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGCATCCGTGG-1</th>\n",
       "      <td>1.182596</td>\n",
       "      <td>1.144585</td>\n",
       "      <td>1.093318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACCTGTCGGAATCT-1</th>\n",
       "      <td>0.982687</td>\n",
       "      <td>1.047557</td>\n",
       "      <td>1.093740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGGTAGTACCT-1</th>\n",
       "      <td>0.417277</td>\n",
       "      <td>0.443068</td>\n",
       "      <td>0.462439</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTCTTATG-1</th>\n",
       "      <td>0.348761</td>\n",
       "      <td>0.282756</td>\n",
       "      <td>0.228207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTGGGCCA-1</th>\n",
       "      <td>2.636122</td>\n",
       "      <td>2.038604</td>\n",
       "      <td>1.597445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAACGGGTCTGTCAAG-1</th>\n",
       "      <td>0.399207</td>\n",
       "      <td>0.446110</td>\n",
       "      <td>0.482377</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGAGAGCTC-1</th>\n",
       "      <td>1.564738</td>\n",
       "      <td>1.329046</td>\n",
       "      <td>1.122280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGAGTACAT-1</th>\n",
       "      <td>1.485496</td>\n",
       "      <td>1.384023</td>\n",
       "      <td>1.281136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGATCCGAG-1</th>\n",
       "      <td>0.699411</td>\n",
       "      <td>0.789952</td>\n",
       "      <td>0.872767</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGAGGGCACTA-1</th>\n",
       "      <td>1.776569</td>\n",
       "      <td>1.709204</td>\n",
       "      <td>1.641042</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGCACTTAACG-1</th>\n",
       "      <td>0.990188</td>\n",
       "      <td>0.850556</td>\n",
       "      <td>0.728069</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGGTCCCTTGT-1</th>\n",
       "      <td>3.447298</td>\n",
       "      <td>3.750008</td>\n",
       "      <td>4.010355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCCGCAAGC-1</th>\n",
       "      <td>0.405894</td>\n",
       "      <td>0.475874</td>\n",
       "      <td>0.539814</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCGCGTTTC-1</th>\n",
       "      <td>1.291190</td>\n",
       "      <td>1.144758</td>\n",
       "      <td>1.010505</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGATGTCGTGGACC-1</th>\n",
       "      <td>3.467186</td>\n",
       "      <td>2.937608</td>\n",
       "      <td>2.486141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGCAAAGAGTACCG-1</th>\n",
       "      <td>2.457092</td>\n",
       "      <td>1.622737</td>\n",
       "      <td>1.104880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGCAATCACTTATC-1</th>\n",
       "      <td>1.255358</td>\n",
       "      <td>1.328561</td>\n",
       "      <td>1.376567</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGAGTTGGC-1</th>\n",
       "      <td>1.755248</td>\n",
       "      <td>1.891669</td>\n",
       "      <td>2.018037</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGCTAACTC-1</th>\n",
       "      <td>1.652005</td>\n",
       "      <td>1.428341</td>\n",
       "      <td>1.230901</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGAGGAGTACC-1</th>\n",
       "      <td>0.543827</td>\n",
       "      <td>0.574121</td>\n",
       "      <td>0.591511</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGCAGTTTACG-1</th>\n",
       "      <td>2.881975</td>\n",
       "      <td>2.747883</td>\n",
       "      <td>2.580114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGCATTAACCG-1</th>\n",
       "      <td>1.117299</td>\n",
       "      <td>0.860455</td>\n",
       "      <td>0.672185</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGGTACTCGCG-1</th>\n",
       "      <td>2.765129</td>\n",
       "      <td>2.744694</td>\n",
       "      <td>2.694334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C1_AAAGTAGTCTCGATGA-1</th>\n",
       "      <td>1.426713</td>\n",
       "      <td>1.493590</td>\n",
       "      <td>1.543274</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCAAACTGCT-1</th>\n",
       "      <td>0.546127</td>\n",
       "      <td>0.668312</td>\n",
       "      <td>0.792676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCACATGACT-1</th>\n",
       "      <td>3.406330</td>\n",
       "      <td>3.133376</td>\n",
       "      <td>2.934748</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCCATACTCTT-1</th>\n",
       "      <td>2.149502</td>\n",
       "      <td>1.922778</td>\n",
       "      <td>1.724100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCGTGGCCCTA-1</th>\n",
       "      <td>2.583077</td>\n",
       "      <td>2.744961</td>\n",
       "      <td>2.847056</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCGTTTAAGCC-1</th>\n",
       "      <td>3.267880</td>\n",
       "      <td>2.653814</td>\n",
       "      <td>2.172043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCAATAAGG-1</th>\n",
       "      <td>0.681796</td>\n",
       "      <td>0.808452</td>\n",
       "      <td>0.928986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCATTATCC-1</th>\n",
       "      <td>3.685506</td>\n",
       "      <td>2.870278</td>\n",
       "      <td>2.267717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCCTTGACC-1</th>\n",
       "      <td>0.337586</td>\n",
       "      <td>0.440907</td>\n",
       "      <td>0.554888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCCTTTCTC-1</th>\n",
       "      <td>2.856774</td>\n",
       "      <td>2.429800</td>\n",
       "      <td>2.113878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTATGCTCTTGGGTA-1</th>\n",
       "      <td>0.425422</td>\n",
       "      <td>0.620381</td>\n",
       "      <td>0.875085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCCATCGATTG-1</th>\n",
       "      <td>0.267748</td>\n",
       "      <td>0.358732</td>\n",
       "      <td>0.460588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCGTAGTACCT-1</th>\n",
       "      <td>2.577453</td>\n",
       "      <td>2.360288</td>\n",
       "      <td>2.148033</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCGTTATGCGT-1</th>\n",
       "      <td>0.959450</td>\n",
       "      <td>1.076300</td>\n",
       "      <td>1.177374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCCTTAATC-1</th>\n",
       "      <td>3.275354</td>\n",
       "      <td>3.232358</td>\n",
       "      <td>3.133913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCGATGAGG-1</th>\n",
       "      <td>0.470076</td>\n",
       "      <td>0.660028</td>\n",
       "      <td>0.894440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTCCTCTCTGTTTGT-1</th>\n",
       "      <td>2.642529</td>\n",
       "      <td>2.428539</td>\n",
       "      <td>2.226831</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGCGCTCTTACCGC-1</th>\n",
       "      <td>0.257352</td>\n",
       "      <td>0.273753</td>\n",
       "      <td>0.284005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTAGCGAGAAA-1</th>\n",
       "      <td>0.294904</td>\n",
       "      <td>0.404630</td>\n",
       "      <td>0.530570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTAGGACACCA-1</th>\n",
       "      <td>0.512239</td>\n",
       "      <td>0.583090</td>\n",
       "      <td>0.645941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTCAAAGGTGC-1</th>\n",
       "      <td>1.495061</td>\n",
       "      <td>1.318807</td>\n",
       "      <td>1.158094</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTGTGATGATA-1</th>\n",
       "      <td>2.325406</td>\n",
       "      <td>2.188960</td>\n",
       "      <td>2.083658</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTGTTGAACTC-1</th>\n",
       "      <td>0.321137</td>\n",
       "      <td>0.383881</td>\n",
       "      <td>0.442784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTTCGTACCGG-1</th>\n",
       "      <td>1.524528</td>\n",
       "      <td>1.443170</td>\n",
       "      <td>1.351901</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGGTTTCTAACGGT-1</th>\n",
       "      <td>0.792568</td>\n",
       "      <td>0.755125</td>\n",
       "      <td>0.708005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAAGTGTCCCG-1</th>\n",
       "      <td>0.841414</td>\n",
       "      <td>0.897749</td>\n",
       "      <td>0.937937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAAGTTAGCGG-1</th>\n",
       "      <td>0.338531</td>\n",
       "      <td>0.370486</td>\n",
       "      <td>0.397700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCACAGTGGAGT-1</th>\n",
       "      <td>0.582584</td>\n",
       "      <td>0.609164</td>\n",
       "      <td>0.624181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCAGTAATTGGA-1</th>\n",
       "      <td>0.254926</td>\n",
       "      <td>0.296702</td>\n",
       "      <td>0.334487</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCATCGTGACAT-1</th>\n",
       "      <td>0.594715</td>\n",
       "      <td>0.876372</td>\n",
       "      <td>1.247693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D7_Dox_C2_TTTGTCATCTCCCTGA-1</th>\n",
       "      <td>1.198471</td>\n",
       "      <td>1.539571</td>\n",
       "      <td>1.896885</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6507 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    g0        g1        g2\n",
       "index                                                     \n",
       "D7_Dox_C1_AAACCTGAGGAATCGC-1  1.268895  0.977589  0.771923\n",
       "D7_Dox_C1_AAACCTGAGTAGATGT-1  2.978660  2.510199  2.130938\n",
       "D7_Dox_C1_AAACCTGCAACCGCCA-1  1.871918  1.789104  1.692330\n",
       "D7_Dox_C1_AAACCTGCACTATCTT-1  3.072989  2.790364  2.516863\n",
       "D7_Dox_C1_AAACCTGCAGCCTGTG-1  1.026225  0.918584  0.815139\n",
       "D7_Dox_C1_AAACCTGCATATACGC-1  1.412582  1.366065  1.303655\n",
       "D7_Dox_C1_AAACCTGCATCCGTGG-1  1.182596  1.144585  1.093318\n",
       "D7_Dox_C1_AAACCTGTCGGAATCT-1  0.982687  1.047557  1.093740\n",
       "D7_Dox_C1_AAACGGGGTAGTACCT-1  0.417277  0.443068  0.462439\n",
       "D7_Dox_C1_AAACGGGTCTCTTATG-1  0.348761  0.282756  0.228207\n",
       "D7_Dox_C1_AAACGGGTCTGGGCCA-1  2.636122  2.038604  1.597445\n",
       "D7_Dox_C1_AAACGGGTCTGTCAAG-1  0.399207  0.446110  0.482377\n",
       "D7_Dox_C1_AAAGATGAGAGAGCTC-1  1.564738  1.329046  1.122280\n",
       "D7_Dox_C1_AAAGATGAGAGTACAT-1  1.485496  1.384023  1.281136\n",
       "D7_Dox_C1_AAAGATGAGATCCGAG-1  0.699411  0.789952  0.872767\n",
       "D7_Dox_C1_AAAGATGAGGGCACTA-1  1.776569  1.709204  1.641042\n",
       "D7_Dox_C1_AAAGATGCACTTAACG-1  0.990188  0.850556  0.728069\n",
       "D7_Dox_C1_AAAGATGGTCCCTTGT-1  3.447298  3.750008  4.010355\n",
       "D7_Dox_C1_AAAGATGTCCGCAAGC-1  0.405894  0.475874  0.539814\n",
       "D7_Dox_C1_AAAGATGTCGCGTTTC-1  1.291190  1.144758  1.010505\n",
       "D7_Dox_C1_AAAGATGTCGTGGACC-1  3.467186  2.937608  2.486141\n",
       "D7_Dox_C1_AAAGCAAAGAGTACCG-1  2.457092  1.622737  1.104880\n",
       "D7_Dox_C1_AAAGCAATCACTTATC-1  1.255358  1.328561  1.376567\n",
       "D7_Dox_C1_AAAGTAGAGAGTTGGC-1  1.755248  1.891669  2.018037\n",
       "D7_Dox_C1_AAAGTAGAGCTAACTC-1  1.652005  1.428341  1.230901\n",
       "D7_Dox_C1_AAAGTAGAGGAGTACC-1  0.543827  0.574121  0.591511\n",
       "D7_Dox_C1_AAAGTAGCAGTTTACG-1  2.881975  2.747883  2.580114\n",
       "D7_Dox_C1_AAAGTAGCATTAACCG-1  1.117299  0.860455  0.672185\n",
       "D7_Dox_C1_AAAGTAGGTACTCGCG-1  2.765129  2.744694  2.694334\n",
       "D7_Dox_C1_AAAGTAGTCTCGATGA-1  1.426713  1.493590  1.543274\n",
       "...                                ...       ...       ...\n",
       "D7_Dox_C2_TTTATGCCAAACTGCT-1  0.546127  0.668312  0.792676\n",
       "D7_Dox_C2_TTTATGCCACATGACT-1  3.406330  3.133376  2.934748\n",
       "D7_Dox_C2_TTTATGCCATACTCTT-1  2.149502  1.922778  1.724100\n",
       "D7_Dox_C2_TTTATGCGTGGCCCTA-1  2.583077  2.744961  2.847056\n",
       "D7_Dox_C2_TTTATGCGTTTAAGCC-1  3.267880  2.653814  2.172043\n",
       "D7_Dox_C2_TTTATGCTCAATAAGG-1  0.681796  0.808452  0.928986\n",
       "D7_Dox_C2_TTTATGCTCATTATCC-1  3.685506  2.870278  2.267717\n",
       "D7_Dox_C2_TTTATGCTCCTTGACC-1  0.337586  0.440907  0.554888\n",
       "D7_Dox_C2_TTTATGCTCCTTTCTC-1  2.856774  2.429800  2.113878\n",
       "D7_Dox_C2_TTTATGCTCTTGGGTA-1  0.425422  0.620381  0.875085\n",
       "D7_Dox_C2_TTTCCTCCATCGATTG-1  0.267748  0.358732  0.460588\n",
       "D7_Dox_C2_TTTCCTCGTAGTACCT-1  2.577453  2.360288  2.148033\n",
       "D7_Dox_C2_TTTCCTCGTTATGCGT-1  0.959450  1.076300  1.177374\n",
       "D7_Dox_C2_TTTCCTCTCCTTAATC-1  3.275354  3.232358  3.133913\n",
       "D7_Dox_C2_TTTCCTCTCGATGAGG-1  0.470076  0.660028  0.894440\n",
       "D7_Dox_C2_TTTCCTCTCTGTTTGT-1  2.642529  2.428539  2.226831\n",
       "D7_Dox_C2_TTTGCGCTCTTACCGC-1  0.257352  0.273753  0.284005\n",
       "D7_Dox_C2_TTTGGTTAGCGAGAAA-1  0.294904  0.404630  0.530570\n",
       "D7_Dox_C2_TTTGGTTAGGACACCA-1  0.512239  0.583090  0.645941\n",
       "D7_Dox_C2_TTTGGTTCAAAGGTGC-1  1.495061  1.318807  1.158094\n",
       "D7_Dox_C2_TTTGGTTGTGATGATA-1  2.325406  2.188960  2.083658\n",
       "D7_Dox_C2_TTTGGTTGTTGAACTC-1  0.321137  0.383881  0.442784\n",
       "D7_Dox_C2_TTTGGTTTCGTACCGG-1  1.524528  1.443170  1.351901\n",
       "D7_Dox_C2_TTTGGTTTCTAACGGT-1  0.792568  0.755125  0.708005\n",
       "D7_Dox_C2_TTTGTCAAGTGTCCCG-1  0.841414  0.897749  0.937937\n",
       "D7_Dox_C2_TTTGTCAAGTTAGCGG-1  0.338531  0.370486  0.397700\n",
       "D7_Dox_C2_TTTGTCACAGTGGAGT-1  0.582584  0.609164  0.624181\n",
       "D7_Dox_C2_TTTGTCAGTAATTGGA-1  0.254926  0.296702  0.334487\n",
       "D7_Dox_C2_TTTGTCATCGTGACAT-1  0.594715  0.876372  1.247693\n",
       "D7_Dox_C2_TTTGTCATCTCCCTGA-1  1.198471  1.539571  1.896885\n",
       "\n",
       "[6507 rows x 3 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# row annotations include initial cell growth rates g0, and also g1 and g2\n",
    "tmap_anno_gr2.obs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tuning $\\lambda_1$ and $\\lambda_2$\n",
    "\n",
    "The numerical value of $\\lambda_1$ determines how different $g_{k+1}$ can be from $g_k$. If growth rates are known very well, then a large value of $\\lambda_1$ can be used to enforce the row-sum constraints strictly and we could just do one growth iteration. When we do more than one growth iteration, the value of $\\lambda_1$ can be interpreted intuitively as an *inverse step size* in our update of growth rates. A large value allows us to change growth by a lot at each iteration. \n",
    "\n",
    "You may have noticed that we set $\\lambda_2 = 50$, a much larger numerical value than $\\lambda_1 = 1$. This enforces the constraints on the column sums strictly (the amount of mass going into each cell $y$ at time $t_2$ is well specified). To see this, we plot a histogram of column sums, and we see a very narrow band. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([   2.,    2.,   23.,   58.,  200.,  624., 1517., 1869.,  726.,\n",
       "          40.]),\n",
       " array([1.404813 , 1.407358 , 1.4099029, 1.4124479, 1.4149928, 1.4175378,\n",
       "        1.4200828, 1.4226277, 1.4251727, 1.4277176, 1.4302626],\n",
       "       dtype=float32),\n",
       " <a list of 10 Patch objects>)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "colsums = tmap_anno_gr2.X.sum(axis=0);\n",
    "plt.hist(colsums)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 545,
     "status": "ok",
     "timestamp": 1556313222255,
     "user": {
      "displayName": "Joshua Gould",
      "photoUrl": "",
      "userId": "09978885733763067346"
     },
     "user_tz": 240
    },
    "id": "P-ZQVITQe9m-",
    "outputId": "a864cea1-48cd-468b-b9ac-845724230979",
    "pycharm": {}
   },
   "source": [
    "# Tuning $\\epsilon$ \n",
    "\n",
    "The parameter $\\epsilon$ controls the level of entropy in the transport maps. A larger value promotes more entropic descendant distributions, so each cell can transition to a larger variety of descendants. We recommend sticking with the default value of $\\epsilon = 0.05$. Note that we rescale the cost matrix to have a median value of $1,$ so the default value of $\\epsilon$ should give similar results for datasets of different scales.  \n",
    "\n",
    "# Compute all transport maps in batch mode\n",
    "\n",
    "The following command computes all transport maps. Note that this can take a few hours on a laptop. For convenience, we include precomputed transport maps in the `tmaps` directory. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 164
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 370,
     "status": "error",
     "timestamp": 1556369739325,
     "user": {
      "displayName": "Joshua Gould",
      "photoUrl": "",
      "userId": "09978885733763067346"
     },
     "user_tz": 240
    },
    "id": "5lFF4FfhAexV",
    "outputId": "2c0fc345-a952-4619-fe70-8a5dcc1a7538",
    "pycharm": {}
   },
   "outputs": [],
   "source": [
    "# ot_model = wot.ot.OTModel(adata, epsilon = 0.05, lambda1 = 1, lambda2 = 50, growth_iters = 3) \n",
    "# ot_model.compute_all_transport_maps(tmap_out='tmaps/serum')\n",
    "\n",
    "# we can speed this up by supplying g2 and doing 1 growth iter"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This can also be run from the command line as follows:\n",
    "\n",
    "`wot optimal_transport --matrix data/ExprMatrix.var.genes.h5ad --cell_days data/cell_days.txt --cell_filter data/serum_cell_ids.txt --cell_growth_rates data/growth_gs_init.txt --growth_iters 3 --lambda1 1 --lambda2 50 --epsilon 0.05`\n",
    "\n",
    "We use these transport maps in the remaining notebooks. "
   ]
  }
 ],
 "metadata": {
  "colab": {
   "name": "wot-2.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.7.3"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "source": [],
    "metadata": {
     "collapsed": false
    }
   }
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "state": {},
    "version_major": 2,
    "version_minor": 0
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}