From 695d0fcb32004105c8325568b9c43ac714316b25 Mon Sep 17 00:00:00 2001
From: Laurane Deckers <laurane.deckers@epita.fr>
Date: Wed, 12 Mar 2025 16:49:33 +0100
Subject: [PATCH] rendu TP4

---
 BINF2025_TP4.ipynb | 68 +++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 58 insertions(+), 10 deletions(-)

diff --git a/BINF2025_TP4.ipynb b/BINF2025_TP4.ipynb
index be4f132..be5b265 100644
--- a/BINF2025_TP4.ipynb
+++ b/BINF2025_TP4.ipynb
@@ -3,9 +3,7 @@
   "nbformat_minor": 0,
   "metadata": {
     "colab": {
-      "provenance": [],
-      "authorship_tag": "ABX9TyOkaYHMV5kAI2JXjqDKbiVa",
-      "include_colab_link": true
+      "provenance": []
     },
     "kernelspec": {
       "name": "python3",
@@ -19,8 +17,7 @@
     {
       "cell_type": "markdown",
       "metadata": {
-        "id": "view-in-github",
-        "colab_type": "text"
+        "id": "view-in-github"
       },
       "source": [
         "<a href=\"https://colab.research.google.com/github/pparutto/BINF2025_TP4/blob/main/BINF2025_TP4.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
@@ -60,7 +57,7 @@
       "metadata": {
         "id": "A8LhiAAeP9u-"
       },
-      "execution_count": null,
+      "execution_count": 1,
       "outputs": []
     },
     {
@@ -75,6 +72,8 @@
     {
       "cell_type": "code",
       "source": [
+        "import numpy as np\n",
+        "\n",
         "def read_fasta(fname):\n",
         "    #Read a fasta file with multiple entries.\n",
         "    #Returns a list of sequences.\n",
@@ -123,7 +122,7 @@
       "metadata": {
         "id": "30qlDWTmQ47E"
       },
-      "execution_count": null,
+      "execution_count": 3,
       "outputs": []
     },
     {
@@ -184,12 +183,20 @@
     {
       "cell_type": "code",
       "source": [
-        "#Votre code ici"
+        "def mark_columns(A: list, alpha: float) -> list :\n",
+        "  marked = []\n",
+        "  for i in range(len(A)) :\n",
+        "    count = 0\n",
+        "    for char in len(A[i]) :\n",
+        "      count += char == \"-\"\n",
+        "    if count / len(A[i]) < alpha :\n",
+        "      marked.append(i)\n",
+        "  return marked"
       ],
       "metadata": {
         "id": "ikf5pcUtRYvV"
       },
-      "execution_count": null,
+      "execution_count": 4,
       "outputs": []
     },
     {
@@ -222,7 +229,48 @@
     {
       "cell_type": "code",
       "source": [
-        "#Votre code ici"
+        "def build_HMM(A: list, mark: list) :\n",
+        "  # Step 1: init matrixes\n",
+        "  N = len(mark)\n",
+        "  T = np.ones(shape=(N+1, 9))\n",
+        "  eM = np.ones(shape=(N+1, 20))\n",
+        "  eI = np.ones(shape=(N+1, 20))\n",
+        "  T[:,0] = 0\n",
+        "  T[:,T.shape[1] - 1] = 0\n",
+        "  eM[:,0] = 0\n",
+        "\n",
+        "  # Step 2: fill matrixes\n",
+        "  pi = []\n",
+        "  for k in range(1, len(A)) :\n",
+        "    for l in range(1, len(A[k])):\n",
+        "      if l in mark :\n",
+        "        if A[k][l] == \"-\" :\n",
+        "          pi[l] = \"D\"\n",
+        "        else :\n",
+        "          pi[l] = \"M\"\n",
+        "      else :\n",
+        "        if A[k][l] == \"-\" :\n",
+        "          pi[l] = \"0\"\n",
+        "        else :\n",
+        "          pi[l] = \"I\"\n",
+        "\n",
+        "  l0 = 0\n",
+        "  while l0 < len(pi) and A[k][l0] == \"-\" :\n",
+        "    l0 += 1\n",
+        "  if l0 == len(pi) :\n",
+        "    eI[0][A[k][l0]]\n",
+        "  T[0][M] += A[k][l0]\n",
+        "\n",
+        "  u = 0\n",
+        "  for l in range(l0, len(pi)) :\n",
+        "    if l in mark :\n",
+        "      u += 1\n",
+        "    elif pi[l] == \"0\" :\n",
+        "      continue\n",
+        "    elif pi[l] == \"M\" and A[k][l] != \"-\" :\n",
+        "      eM[u][A[k][l]] += 1\n",
+        "    elif pi[l] == \"I\" and A[k][l] != \"-\" :\n",
+        "      eI[u][A[k][l]] += 1"
       ],
       "metadata": {
         "id": "Y3TXPqNhTwNo"