Move from alpharaw

docs/build_docs.sh

@@ -1,11 +1,9 @@
 rm -rf build
 conda env remove -n alphabasedocs
 conda create -n alphabasedocs python=3.9 -y
-# conda create -n alphatimsinstaller python=3.9
+
 conda activate alphabasedocs
-# call conda install git -y
-# call pip install 'git+https://github.com/MannLabs/alphatims.git#egg=alphatims[gui]' --use-feature=2020-resolver
-# brew install freetype
+
 pip install '../.[development]'
 make html
 conda deactivate

nbs_tests/match/psm_match.ipynb

@@ -0,0 +1,333 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#---#| default_exp match.psm_match"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Match"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Peak matching functionalities"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "\n",
+    "from peptdeep.match.psm_match import PepSpecMatch\n",
+    "from alpharaw import register_all_readers\n",
+    "from alpharaw.ms_data_base import ms_reader_provider\n",
+    "\n",
+    "register_all_readers()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#| hide\n",
+    "import io\n",
+    "import copy\n",
+    "\n",
+    "from alphabase.psm_reader.pfind_reader import pFindReader\n",
+    "from alphabase.peptide.fragment import create_fragment_mz_dataframe_by_sort_precursor\n",
+    "from alpharaw.legacy_msdata.mgf import MGFReader"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 2/2 [00:01<00:00,  1.41it/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "#| hide\n",
+    "#unittest\n",
+    "mgf = io.StringIO(\"\"\"\n",
+    "BEGIN IONS\n",
+    "TITLE=02445a_BA7-TUM_HLA_7_01_01-DDA-1h-R1.31809.31809.3.0.dta\n",
+    "CHARGE=3+\n",
+    "RTINSECONDS=0.5418930\n",
+    "PEPMASS=272.276336\n",
+    "103.92207 5457.3\n",
+    "104.20045 5051.4\n",
+    "108.70090 5891.7\n",
+    "113.94175 6442.6\n",
+    "116.92975 40506.3\n",
+    "116.93716 8945.5\n",
+    "128.37773 6427.8\n",
+    "131.95308 288352.6\n",
+    "133.93259 7344.6\n",
+    "138.44611 7326.1\n",
+    "139.00072 41556.8\n",
+    "140.00319 16738.8\n",
+    "140.99719 9493.8\n",
+    "145.93156 10209.3\n",
+    "145.94897 10497.8\n",
+    "147.94559 8206.3\n",
+    "147.96396 30552.8\n",
+    "148.95543 14654.7\n",
+    "149.96338 234207.8\n",
+    "150.95096 8306.0\n",
+    "157.01089 84638.9\n",
+    "158.01357 27925.7\n",
+    "159.00627 16084.7\n",
+    "163.94281 24751.1\n",
+    "163.95915 32203.3\n",
+    "165.95605 44458.0\n",
+    "165.97186 11530.2\n",
+    "166.99500 26432.2\n",
+    "167.97302 9216.7\n",
+    "181.95230 13858.8\n",
+    "191.95448 66152.7\n",
+    "192.95538 8408.9\n",
+    "193.07185 9092.8\n",
+    "193.95313 660574.9\n",
+    "194.95674 23452.8\n",
+    "194.99008 143940.9\n",
+    "200.00568 19510.8\n",
+    "200.99942 23678.7\n",
+    "204.30894 9406.1\n",
+    "209.96466 21853.6\n",
+    "211.96245 65351.0\n",
+    "218.90355 9149.6\n",
+    "223.91072 11300.2\n",
+    "238.89684 12108.8\n",
+    "243.93825 10150.2\n",
+    "243.97040 10987.7\n",
+    "244.94121 8744.2\n",
+    "246.90314 11556.3\n",
+    "271.93225 29430.0\n",
+    "271.99219 51184.4\n",
+    "272.19150 31960.4\n",
+    "272.98602 35844.1\n",
+    "273.94431 11031.8\n",
+    "284.47998 8191.3\n",
+    "290.00125 66212.4\n",
+    "290.99539 54064.7\n",
+    "293.89490 10005.0\n",
+    "407.06372 10838.2\n",
+    "464.36697 9715.4\n",
+    "633.40036 633.40036\n",
+    "698.81390 9711.7\n",
+    "707.301117 707.301117\n",
+    "END IONS\n",
+    "BEGIN IONS\n",
+    "TITLE=02445a_BA7-TUM_HLA_7_01_01-DDA-1h-R1.23862.23862.2.0.dta\n",
+    "CHARGE=2+\n",
+    "RTINSECONDS=0.6455220\n",
+    "PEPMASS=287.427959\n",
+    "103.34669 5304.0\n",
+    "104.66884 5639.7\n",
+    "113.42419 6258.3\n",
+    "118.84039 5837.5\n",
+    "119.93203 13977.3\n",
+    "130.69589 6876.2\n",
+    "133.94824 43094.3\n",
+    "134.30524 7671.5\n",
+    "135.96359 9031.3\n",
+    "138.99994 8329.7\n",
+    "146.95573 31143.9\n",
+    "147.96323 12176.5\n",
+    "150.95151 65859.3\n",
+    "151.95818 24384.2\n",
+    "157.01105 19241.5\n",
+    "157.34985 7532.5\n",
+    "161.08838 7843.9\n",
+    "161.94234 20119.7\n",
+    "162.95146 60110.4\n",
+    "163.95877 183305.5\n",
+    "164.96657 13647.5\n",
+    "174.95139 150331.9\n",
+    "175.95258 21393.4\n",
+    "178.94460 11433.1\n",
+    "179.95316 13650.5\n",
+    "180.96204 15353.5\n",
+    "190.94572 30418.9\n",
+    "191.95422 61914.1\n",
+    "192.61461 8642.1\n",
+    "192.94395 12331.4\n",
+    "192.96207 132342.5\n",
+    "193.96318 19303.0\n",
+    "209.04164 25149.6\n",
+    "209.96368 154185.0\n",
+    "209.98361 12353.5\n",
+    "213.86244 11541.3\n",
+    "224.93071 12903.0\n",
+    "228.92879 8773.6\n",
+    "241.86043 135357.5\n",
+    "242.86113 20805.2\n",
+    "242.94327 26679.4\n",
+    "243.95219 29569.9\n",
+    "244.92361 12153.5\n",
+    "246.90300 16650.3\n",
+    "252.96521 73484.3\n",
+    "253.96646 11527.5\n",
+    "286.85858 10166.4\n",
+    "287.94186 18763.2\n",
+    "303.87665 39189.3\n",
+    "304.88116 11976.0\n",
+    "321.89087 97122.5\n",
+    "322.88867 28020.8\n",
+    "370.28696 9008.2\n",
+    "389.82578 13277.0\n",
+    "407.83545 12220.4\n",
+    "425.84872 13236.5\n",
+    "482.54852 10940.2\n",
+    "END IONS\n",
+    "BEGIN IONS\n",
+    "TITLE=02445a_BA7-TUM_HLA_7_01_01-DDA-1h-R1.23431.23431.2.0.dta\n",
+    "CHARGE=2+\n",
+    "RTINSECONDS=0.6455220\n",
+    "PEPMASS=287.427959\n",
+    "103.34669 5304.0\n",
+    "104.66884 5639.7\n",
+    "END IONS\n",
+    "BEGIN IONS\n",
+    "TITLE=02445a_BA7-TUM_HLA_7_01_01-DDA-1h-R1.32733.32733.2.0.dta\n",
+    "CHARGE=2+\n",
+    "RTINSECONDS=0.6455220\n",
+    "PEPMASS=287.427959\n",
+    "103.34669 5304.0\n",
+    "104.66884 5639.7\n",
+    "402.705571 402.705571\n",
+    "END IONS\n",
+    "BEGIN IONS\n",
+    "TITLE=02445a_BA7-TUM_HLA_7_01_01-DDA-1h-R1.23669.23669.2.0.dta\n",
+    "CHARGE=2+\n",
+    "RTINSECONDS=0.6455220\n",
+    "PEPMASS=287.427959\n",
+    "END IONS\n",
+    "\"\"\")\n",
+    "\n",
+    "ms_file_dict = {\n",
+    "    'raw': copy.deepcopy(mgf),\n",
+    "    'raw1': copy.deepcopy(mgf),\n",
+    "}\n",
+    "\n",
+    "psmlabel_str = '''File_Name\tSequence\tModification\tCharge\tScan_No\tProteins\tQ-value\tTarget/Decoy\tFinal_Score\n",
+    "raw.31809.31809.2.0.dta\tPSTDLLMLK\t2,Phospho[S];7,Oxidation[M];\t2\t31809\tProt\t0\ttarget\t100\n",
+    "raw.23862.23862.2.0.dta\tHTAYSDFLSDK\t\t2\t23862\tProt\t0\ttarget\t100\n",
+    "raw.23431.23431.2.0.dta\tHTAYSDFLSDK\t\t2\t23431\tProt\t0\ttarget\t100\n",
+    "raw.32733.32733.2.0.dta\tHFALFSTDVTK\t\t2\t32733\tProt\t0\ttarget\t100\n",
+    "raw.23669.23669.2.0.dta\tHTAYSDFLSDK\t\t2\t23669\tProt\t0\ttarget\t100\n",
+    "raw1.31809.31809.2.0.dta\tPSTDLLMLK\t2,Phospho[S];7,Oxidation[M];\t2\t31809\tProt\t0\ttarget\t100\n",
+    "raw1.23862.23862.2.0.dta\tHTAYSDFLSDK\t\t2\t23862\tProt\t0\ttarget\t100\n",
+    "raw1.23431.23431.2.0.dta\tHTAYSDFLSDK\t\t2\t23431\tProt\t0\ttarget\t100\n",
+    "raw1.32733.32733.2.0.dta\tHFALFSTDVTK\t\t2\t32733\tProt\t0\ttarget\t100\n",
+    "'''\n",
+    "reader = pFindReader()\n",
+    "reader.import_file(io.StringIO(psmlabel_str))\n",
+    "psm_df = reader.psm_df\n",
+    "matching = PepSpecMatch()\n",
+    "matching.match_ms2_multi_raw(psm_df, ms_file_dict, 'mgf')\n",
+    "merrs = matching.matched_mz_err_df.values\n",
+    "#np.sum(matching.matched_intensity_df.values!=0,axis=1)\n",
+    "assert len(merrs[~np.isinf(merrs)])==6\n",
+    "assert np.count_nonzero(matching.matched_intensity_df.values)==6"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 2/2 [00:00<00:00, 174.90it/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "#| hide\n",
+    "mgf_reader = ms_reader_provider.get_reader('mgf')\n",
+    "mgf_reader.import_raw(copy.deepcopy(mgf))\n",
+    "ms_file_dict = {'raw': mgf_reader}\n",
+    "mgf_reader1 = ms_reader_provider.get_reader('mgf')\n",
+    "mgf_reader1.import_raw(copy.deepcopy(mgf))\n",
+    "ms_file_dict['raw1'] = mgf_reader1\n",
+    "matching.match_ms2_multi_raw(psm_df, ms_file_dict, 'mgf')\n",
+    "merrs = matching.matched_mz_err_df.values\n",
+    "assert np.count_nonzero(matching.matched_intensity_df.values) == 6\n",
+    "assert len(merrs[~np.isinf(merrs)]) == 6"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 1/1 [00:00<00:00, 111.73it/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "#| hide\n",
+    "ms_file_dict = {\n",
+    "    'raw': copy.deepcopy(mgf),\n",
+    "}\n",
+    "reader = pFindReader()\n",
+    "reader.import_file(io.StringIO(psmlabel_str))\n",
+    "psm_df = reader.psm_df\n",
+    "psm_df = psm_df[~psm_df.raw_name.str.startswith('raw1')].copy()\n",
+    "matching = PepSpecMatch()\n",
+    "matching.match_ms2_multi_raw(psm_df, ms_file_dict, 'mgf')\n",
+    "matching.load_ms_data(copy.deepcopy(mgf), 'mgf')\n",
+    "df, frag_mz_df, frag_inten_df, frag_merr_df = matching.match_ms2_one_raw(\n",
+    "    psm_df\n",
+    ")\n",
+    "assert (matching.fragment_mz_df==frag_mz_df).values.all()\n",
+    "assert (matching.matched_intensity_df==frag_inten_df).values.all()\n",
+    "assert (matching.matched_mz_err_df==frag_merr_df).values.all()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3.8.3 ('base')",
+   "language": "python",
+   "name": "python3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}