prealignment_sequence_prep.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "from Bio import SeqIO\n",
    "import csv\n",
    "import os\n",
    "import pandas as pd\n",
    "from pyfasta import Fasta"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1011 Genomes S. cerevisiae ORFs (Peter, et al. 2018)\n",
    "\n",
    "Write files for each gene with only strains in a given population"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'AAA': ['1WineEuropean'],\n",
       " 'AAB': ['7Mosaicbeer'],\n",
       " 'AAC': ['11Alebeer'],\n",
       " 'AAD': ['25Sake'],\n",
       " 'AAE': ['1WineEuropeansubclade4'],\n",
       " 'AAG': ['M3Mosaicregion3'],\n",
       " 'AAI': ['1WineEuropean'],\n",
       " 'AAK': ['1WineEuropeansubclade4'],\n",
       " 'AAL': ['8Mixedorigin'],\n",
       " 'AAM': ['8Mixedorigin'],\n",
       " 'AAN': ['8Mixedorigin'],\n",
       " 'AAP': ['8Mixedorigin'],\n",
       " 'AAQ': ['7Mosaicbeer'],\n",
       " 'AAR': ['7Mosaicbeer'],\n",
       " 'AAS': ['M3Mosaicregion3'],\n",
       " 'AAT': ['M3Mosaicregion3'],\n",
       " 'AAV': ['M3Mosaicregion3'],\n",
       " 'ABA': ['M3Mosaicregion3'],\n",
       " 'ABB': ['26Asianfermentation'],\n",
       " 'ABC': ['M3Mosaicregion3'],\n",
       " 'ABD': ['M3Mosaicregion3'],\n",
       " 'ABE': ['1WineEuropean'],\n",
       " 'ABF': ['M3Mosaicregion3'],\n",
       " 'ABG': ['M3Mosaicregion3'],\n",
       " 'ABH': ['M3Mosaicregion3'],\n",
       " 'ABI': ['M3Mosaicregion3'],\n",
       " 'ABK': ['M3Mosaicregion3'],\n",
       " 'ABM': ['2Alpechin'],\n",
       " 'ABP': ['1WineEuropean'],\n",
       " 'ABQ': ['1WineEuropean'],\n",
       " 'ABR': ['1WineEuropean'],\n",
       " 'ABS': ['1WineEuropean'],\n",
       " 'ABT': ['1WineEuropean'],\n",
       " 'ABV': ['8Mixedorigin'],\n",
       " 'ACA': ['M3Mosaicregion3'],\n",
       " 'ACB': ['M3Mosaicregion3'],\n",
       " 'ACC': ['21Ecuadorean'],\n",
       " 'ACD': ['23NorthAmericanoak'],\n",
       " 'ACF': ['1WineEuropean'],\n",
       " 'ACG': ['M1Mosaicregion1'],\n",
       " 'ACH': ['M1Mosaicregion1'],\n",
       " 'ACI': ['M3Mosaicregion3'],\n",
       " 'ACK': ['26Asianfermentation'],\n",
       " 'ACL': ['M3Mosaicregion3'],\n",
       " 'ACM': ['26Asianfermentation'],\n",
       " 'ACN': ['M3Mosaicregion3'],\n",
       " 'ACP': ['1WineEuropeansubclade1'],\n",
       " 'ACQ': ['18FarEastAsia'],\n",
       " 'ACR': ['1WineEuropean'],\n",
       " 'ACS': ['M1Mosaicregion1'],\n",
       " 'ACT': ['1WineEuropean'],\n",
       " 'ACV': ['1WineEuropean'],\n",
       " 'ADA': ['M3Mosaicregion3'],\n",
       " 'ADB': ['1WineEuropean'],\n",
       " 'ADC': ['1WineEuropean'],\n",
       " 'ADD': ['1WineEuropeansubclade2'],\n",
       " 'ADF': ['13Africanpalmwine'],\n",
       " 'ADH': ['26Asianfermentation'],\n",
       " 'ADI': ['1WineEuropeansubclade2'],\n",
       " 'ADK': ['M3Mosaicregion3'],\n",
       " 'ADL': ['8Mixedorigin'],\n",
       " 'ADM': ['8Mixedorigin'],\n",
       " 'ADN': ['25Sake'],\n",
       " 'ADP': ['8Mixedorigin'],\n",
       " 'ADQ': ['25Sake'],\n",
       " 'ADR': ['1WineEuropean'],\n",
       " 'ADS': ['1WineEuropean'],\n",
       " 'ADT': ['M3Mosaicregion3'],\n",
       " 'ADV': ['13Africanpalmwine'],\n",
       " 'AEA': ['7Mosaicbeer'],\n",
       " 'AEB': ['11Alebeer'],\n",
       " 'AEC': ['11Alebeer'],\n",
       " 'AEE': ['M3Mosaicregion3'],\n",
       " 'AEF': ['25Sake'],\n",
       " 'AEG': ['3Brazilianbioethanol'],\n",
       " 'AEH': ['3Brazilianbioethanol'],\n",
       " 'AEI': ['M3Mosaicregion3'],\n",
       " 'AEK': ['1WineEuropeansubclade4'],\n",
       " 'AEL': ['M3Mosaicregion3'],\n",
       " 'AEM': ['8Mixedorigin'],\n",
       " 'AEN': ['6Africanbeer'],\n",
       " 'AEP': ['26Asianfermentation'],\n",
       " 'AEQ': ['7Mosaicbeer'],\n",
       " 'AER': ['M2Mosaicregion2'],\n",
       " 'AES': ['1WineEuropeansubclade4'],\n",
       " 'AET': ['1WineEuropeansubclade4'],\n",
       " 'AEV': ['8Mixedorigin'],\n",
       " 'AFA': ['7Mosaicbeer'],\n",
       " 'AFB': ['7Mosaicbeer'],\n",
       " 'AFC': ['8Mixedorigin'],\n",
       " 'AFD': ['3Brazilianbioethanol'],\n",
       " 'AFE': ['8Mixedorigin'],\n",
       " 'AFF': ['8Mixedorigin'],\n",
       " 'AFG': ['M3Mosaicregion3'],\n",
       " 'AFH': ['26Asianfermentation'],\n",
       " 'AFI': ['1WineEuropeansubclade4'],\n",
       " 'AFK': ['24Asianislands'],\n",
       " 'AFL': ['6Africanbeer'],\n",
       " 'AFM': ['25Sake'],\n",
       " 'AFN': ['25Sake'],\n",
       " 'AFP': ['7Mosaicbeer'],\n",
       " 'AFQ': ['M2Mosaicregion2'],\n",
       " 'AFR': ['3Brazilianbioethanol'],\n",
       " 'AFT': ['8Mixedorigin'],\n",
       " 'AFV': ['1WineEuropeansubclade4'],\n",
       " 'AGA': ['1WineEuropean'],\n",
       " 'AGB': ['11Alebeer'],\n",
       " 'AGE': ['8Mixedorigin'],\n",
       " 'AGF': ['1WineEuropean'],\n",
       " 'AGG': ['26Asianfermentation'],\n",
       " 'AGH': ['1WineEuropean'],\n",
       " 'AGI': ['1WineEuropean'],\n",
       " 'AGK': ['8Mixedorigin'],\n",
       " 'AGL': ['1WineEuropean'],\n",
       " 'AGM': ['3Brazilianbioethanol'],\n",
       " 'AGN': ['M1Mosaicregion1'],\n",
       " 'AGP': ['M3Mosaicregion3'],\n",
       " 'AGR': ['24Asianislands'],\n",
       " 'AGS': ['1WineEuropean'],\n",
       " 'AGT': ['M3Mosaicregion3'],\n",
       " 'AGV': ['1WineEuropean'],\n",
       " 'AHA': ['M3Mosaicregion3'],\n",
       " 'AHC': ['3Brazilianbioethanol'],\n",
       " 'AHD': ['1WineEuropean'],\n",
       " 'AHE': ['1WineEuropean'],\n",
       " 'AHF': ['M1Mosaicregion1'],\n",
       " 'AHG': ['1WineEuropean'],\n",
       " 'AHH': ['24Asianislands'],\n",
       " 'AHI': ['1WineEuropean'],\n",
       " 'AHK': ['1WineEuropeansubclade4'],\n",
       " 'AHL': ['2Alpechin'],\n",
       " 'AHM': ['2Alpechin'],\n",
       " 'AHN': ['1WineEuropean'],\n",
       " 'AHP': ['1WineEuropeansubclade4'],\n",
       " 'AHQ': ['26Asianfermentation'],\n",
       " 'AHR': ['1WineEuropean'],\n",
       " 'AHS': ['13Africanpalmwine'],\n",
       " 'AHT': ['1WineEuropean'],\n",
       " 'AHV': ['8Mixedorigin'],\n",
       " 'AIA': ['1WineEuropeansubclade4'],\n",
       " 'AIB': ['1WineEuropeansubclade4'],\n",
       " 'AIC': ['1WineEuropeansubclade4'],\n",
       " 'AID': ['2Alpechin'],\n",
       " 'AIE': ['2Alpechin'],\n",
       " 'AIF': ['1WineEuropean'],\n",
       " 'AIG': ['7Mosaicbeer'],\n",
       " 'AII': ['1WineEuropean'],\n",
       " 'AIK': ['1WineEuropean'],\n",
       " 'AIL': ['1WineEuropean'],\n",
       " 'AIM': ['1WineEuropean'],\n",
       " 'AIN': ['1WineEuropean'],\n",
       " 'AIP': ['1WineEuropeansubclade4'],\n",
       " 'AIQ': ['1WineEuropean'],\n",
       " 'AIR': ['1WineEuropean'],\n",
       " 'AIT': ['1WineEuropean'],\n",
       " 'AKB': ['13Africanpalmwine'],\n",
       " 'AKC': ['13Africanpalmwine'],\n",
       " 'AKD': ['13Africanpalmwine'],\n",
       " 'AKE': ['13Africanpalmwine'],\n",
       " 'AKG': ['13Africanpalmwine'],\n",
       " 'AKH': ['13Africanpalmwine'],\n",
       " 'AKI': ['M3Mosaicregion3'],\n",
       " 'AKL': ['26Asianfermentation'],\n",
       " 'AKM': ['19Malaysian'],\n",
       " 'AKN': ['23NorthAmericanoak'],\n",
       " 'AKP': ['1WineEuropean'],\n",
       " 'AKQ': ['M3Mosaicregion3'],\n",
       " 'AKR': ['26Asianfermentation'],\n",
       " 'AKS': ['M3Mosaicregion3'],\n",
       " 'AKT': ['8Mixedorigin'],\n",
       " 'AKV': ['1WineEuropean'],\n",
       " 'ALA': ['26Asianfermentation'],\n",
       " 'ALB': ['26Asianfermentation'],\n",
       " 'ALC': ['26Asianfermentation'],\n",
       " 'ALD': ['5Frenchdairy'],\n",
       " 'ALE': ['1WineEuropean'],\n",
       " 'ALG': ['M3Mosaicregion3'],\n",
       " 'ALH': ['26Asianfermentation'],\n",
       " 'ALK': ['26Asianfermentation'],\n",
       " 'ALL': ['1WineEuropean'],\n",
       " 'ALM': ['1WineEuropean'],\n",
       " 'ALN': ['8Mixedorigin'],\n",
       " 'ALP': ['8Mixedorigin'],\n",
       " 'ALQ': ['1WineEuropeansubclade1'],\n",
       " 'ALR': ['M3Mosaicregion3'],\n",
       " 'ALS': ['1WineEuropeansubclade1'],\n",
       " 'ALT': ['1WineEuropeansubclade1'],\n",
       " 'ALV': ['8Mixedorigin'],\n",
       " 'AMA': ['1WineEuropeansubclade1'],\n",
       " 'AMG': ['8Mixedorigin'],\n",
       " 'AMH': ['17Taiwanese'],\n",
       " 'AMI': ['8Mixedorigin'],\n",
       " 'AMK': ['8Mixedorigin'],\n",
       " 'AML': ['8Mixedorigin'],\n",
       " 'AMN': ['8Mixedorigin'],\n",
       " 'AMP': ['M1Mosaicregion1'],\n",
       " 'AMQ': ['8Mixedorigin'],\n",
       " 'AMR': ['1WineEuropean'],\n",
       " 'AMS': ['1WineEuropean'],\n",
       " 'AMT': ['1WineEuropeansubclade3'],\n",
       " 'ANA': ['M3Mosaicregion3'],\n",
       " 'ANC': ['23NorthAmericanoak'],\n",
       " 'AND': ['23NorthAmericanoak'],\n",
       " 'ANE': ['23NorthAmericanoak'],\n",
       " 'ANF': ['23NorthAmericanoak'],\n",
       " 'ANG': ['23NorthAmericanoak'],\n",
       " 'ANH': ['23NorthAmericanoak'],\n",
       " 'ANI': ['23NorthAmericanoak'],\n",
       " 'ANK': ['23NorthAmericanoak'],\n",
       " 'ANL': ['6Africanbeer'],\n",
       " 'ANM': ['6Africanbeer'],\n",
       " 'ANN': ['12WestAfricancocoa'],\n",
       " 'ANP': ['8Mixedorigin'],\n",
       " 'ANQ': ['1WineEuropean'],\n",
       " 'ANR': ['25Sake'],\n",
       " 'ANS': ['25Sake'],\n",
       " 'ANT': ['25Sake'],\n",
       " 'ANV': ['25Sake'],\n",
       " 'APA': ['25Sake'],\n",
       " 'APC': ['25Sake'],\n",
       " 'APD': ['25Sake'],\n",
       " 'APE': ['25Sake'],\n",
       " 'APF': ['25Sake'],\n",
       " 'APG': ['3Brazilianbioethanol'],\n",
       " 'APH': ['3Brazilianbioethanol'],\n",
       " 'API': ['8Mixedorigin'],\n",
       " 'APK': ['8Mixedorigin'],\n",
       " 'APL': ['8Mixedorigin'],\n",
       " 'APM': ['8Mixedorigin'],\n",
       " 'APN': ['8Mixedorigin'],\n",
       " 'APP': ['7Mosaicbeer'],\n",
       " 'APQ': ['13Africanpalmwine'],\n",
       " 'APR': ['13Africanpalmwine'],\n",
       " 'APS': ['1WineEuropean'],\n",
       " 'APT': ['1WineEuropean'],\n",
       " 'APV': ['1WineEuropean'],\n",
       " 'AQA': ['2Alpechin'],\n",
       " 'AQB': ['1WineEuropean'],\n",
       " 'AQC': ['1WineEuropeansubclade4'],\n",
       " 'AQD': ['2Alpechin'],\n",
       " 'AQE': ['1WineEuropeansubclade4'],\n",
       " 'AQG': ['7Mosaicbeer'],\n",
       " 'AQH': ['7Mosaicbeer'],\n",
       " 'AQI': ['8Mixedorigin'],\n",
       " 'AQM': ['5Frenchdairy'],\n",
       " 'AQN': ['25Sake'],\n",
       " 'AQP': ['1WineEuropeansubclade4'],\n",
       " 'AQQ': ['1WineEuropean'],\n",
       " 'AQR': ['1WineEuropean'],\n",
       " 'AQS': ['26Asianfermentation'],\n",
       " 'AQT': ['11Alebeer'],\n",
       " 'ARA': ['M3Mosaicregion3'],\n",
       " 'ARB': ['M3Mosaicregion3'],\n",
       " 'ARC': ['1WineEuropean'],\n",
       " 'ARE': ['7Mosaicbeer'],\n",
       " 'ARG': ['24Asianislands'],\n",
       " 'ARH': ['24Asianislands'],\n",
       " 'ARI': ['24Asianislands'],\n",
       " 'ARK': ['1WineEuropean'],\n",
       " 'ARL': ['1WineEuropean'],\n",
       " 'ARM': ['25Sake'],\n",
       " 'ARN': ['5Frenchdairy'],\n",
       " 'ARP': ['25Sake'],\n",
       " 'ARQ': ['1WineEuropean'],\n",
       " 'ARR': ['5Frenchdairy'],\n",
       " 'ARS': ['5Frenchdairy'],\n",
       " 'ART': ['1WineEuropeansubclade4'],\n",
       " 'ARV': ['1WineEuropean'],\n",
       " 'ASA': ['1WineEuropeansubclade4'],\n",
       " 'ASB': ['7Mosaicbeer'],\n",
       " 'ASC': ['6Africanbeer'],\n",
       " 'ASD': ['1WineEuropean'],\n",
       " 'ASE': ['1WineEuropean'],\n",
       " 'ASG': ['1WineEuropean'],\n",
       " 'ASH': ['11Alebeer'],\n",
       " 'ASL': ['8Mixedorigin'],\n",
       " 'ASN': ['5Frenchdairy'],\n",
       " 'ASP': ['8Mixedorigin'],\n",
       " 'ASR': ['1WineEuropean'],\n",
       " 'ASS': ['1WineEuropean'],\n",
       " 'AST': ['M1Mosaicregion1'],\n",
       " 'ASV': ['M1Mosaicregion1'],\n",
       " 'ATA': ['1WineEuropean'],\n",
       " 'ATB': ['M3Mosaicregion3'],\n",
       " 'ATC': ['M3Mosaicregion3'],\n",
       " 'ATD': ['M3Mosaicregion3'],\n",
       " 'ATE': ['M3Mosaicregion3'],\n",
       " 'ATH': ['1WineEuropean'],\n",
       " 'ATM': ['1WineEuropean'],\n",
       " 'ATN': ['1WineEuropean'],\n",
       " 'ATP': ['13Africanpalmwine'],\n",
       " 'ATQ': ['2Alpechin'],\n",
       " 'ATR': ['8Mixedorigin'],\n",
       " 'ATS': ['8Mixedorigin'],\n",
       " 'ATV': ['11Alebeer'],\n",
       " 'AVB': ['2Alpechin'],\n",
       " 'AVC': ['25Sake'],\n",
       " 'AVD': ['25Sake'],\n",
       " 'AVH': ['1WineEuropean'],\n",
       " 'AVI': ['23NorthAmericanoak'],\n",
       " 'AVK': ['1WineEuropean'],\n",
       " 'AVL': ['25Sake'],\n",
       " 'AVM': ['25Sake'],\n",
       " 'AVN': ['6Africanbeer'],\n",
       " 'AVP': ['6Africanbeer'],\n",
       " 'AVQ': ['6Africanbeer'],\n",
       " 'AVR': ['6Africanbeer'],\n",
       " 'AVS': ['6Africanbeer'],\n",
       " 'AVT': ['6Africanbeer'],\n",
       " 'AVV': ['13Africanpalmwine'],\n",
       " 'BAA': ['13Africanpalmwine'],\n",
       " 'BAB': ['13Africanpalmwine'],\n",
       " 'BAC': ['13Africanpalmwine'],\n",
       " 'BAD': ['13Africanpalmwine'],\n",
       " 'BAE': ['13Africanpalmwine'],\n",
       " 'BAF': ['13Africanpalmwine'],\n",
       " 'BAG': ['15CHNII'],\n",
       " 'BAH': ['15CHNII'],\n",
       " 'BAI': ['18FarEastAsia'],\n",
       " 'BAK': ['18FarEastAsia'],\n",
       " 'BAL': ['16CHNI'],\n",
       " 'BAM': ['14CHNIII'],\n",
       " 'BAN': ['20CHNV'],\n",
       " 'BAP': ['20CHNV'],\n",
       " 'BAQ': ['14CHNIII'],\n",
       " 'BAR': ['1WineEuropean'],\n",
       " 'BAS': ['1WineEuropean'],\n",
       " 'BAT': ['1WineEuropean'],\n",
       " 'BAV': ['1WineEuropean'],\n",
       " 'BBA': ['1WineEuropean'],\n",
       " 'BBB': ['1WineEuropean'],\n",
       " 'BBC': ['8Mixedorigin'],\n",
       " 'BBD': ['8Mixedorigin'],\n",
       " 'BBE': ['8Mixedorigin'],\n",
       " 'BBF': ['8Mixedorigin'],\n",
       " 'BBG': ['11Alebeer'],\n",
       " 'BBH': ['1WineEuropeansubclade1'],\n",
       " 'BBI': ['1WineEuropeansubclade1'],\n",
       " 'BBK': ['1WineEuropeansubclade1'],\n",
       " 'BBL': ['M3Mosaicregion3'],\n",
       " 'BBP': ['8Mixedorigin'],\n",
       " 'BBQ': ['4Mediterraneanoak'],\n",
       " 'BBS': ['4Mediterraneanoak'],\n",
       " 'BBT': ['M3Mosaicregion3'],\n",
       " 'BBV': ['1WineEuropean'],\n",
       " 'BCA': ['1WineEuropeansubclade3'],\n",
       " 'BCB': ['10FrenchGuianahuman'],\n",
       " 'BCC': ['10FrenchGuianahuman'],\n",
       " 'BCD': ['10FrenchGuianahuman'],\n",
       " 'BCE': ['10FrenchGuianahuman'],\n",
       " 'BCF': ['10FrenchGuianahuman'],\n",
       " 'BCG': ['10FrenchGuianahuman'],\n",
       " 'BCH': ['10FrenchGuianahuman'],\n",
       " 'BCI': ['10FrenchGuianahuman'],\n",
       " 'BCK': ['10FrenchGuianahuman'],\n",
       " 'BCL': ['10FrenchGuianahuman'],\n",
       " 'BCM': ['10FrenchGuianahuman'],\n",
       " 'BCN': ['10FrenchGuianahuman'],\n",
       " 'BCP': ['10FrenchGuianahuman'],\n",
       " 'BCQ': ['10FrenchGuianahuman'],\n",
       " 'BCR': ['8Mixedorigin'],\n",
       " 'BCS': ['8Mixedorigin'],\n",
       " 'BCT': ['8Mixedorigin'],\n",
       " 'BCV': ['M2Mosaicregion2'],\n",
       " 'BDA': ['M2Mosaicregion2'],\n",
       " 'BDB': ['M2Mosaicregion2'],\n",
       " 'BDC': ['M3Mosaicregion3'],\n",
       " 'BDD': ['M2Mosaicregion2'],\n",
       " 'BDE': ['M2Mosaicregion2'],\n",
       " 'BDF': ['M3Mosaicregion3'],\n",
       " 'BDG': ['M3Mosaicregion3'],\n",
       " 'BDH': ['M2Mosaicregion2'],\n",
       " 'BDI': ['M2Mosaicregion2'],\n",
       " 'BDK': ['M2Mosaicregion2'],\n",
       " 'BDL': ['8Mixedorigin'],\n",
       " 'BDM': ['M3Mosaicregion3'],\n",
       " 'BDN': ['6Africanbeer'],\n",
       " 'BDP': ['10FrenchGuianahuman'],\n",
       " 'BDQ': ['10FrenchGuianahuman'],\n",
       " 'BDR': ['10FrenchGuianahuman'],\n",
       " 'BDS': ['10FrenchGuianahuman'],\n",
       " 'BDT': ['10FrenchGuianahuman'],\n",
       " 'BDV': ['10FrenchGuianahuman'],\n",
       " 'BEA': ['10FrenchGuianahuman'],\n",
       " 'BEB': ['10FrenchGuianahuman'],\n",
       " 'BEC': ['10FrenchGuianahuman'],\n",
       " 'BED': ['10FrenchGuianahuman'],\n",
       " 'BEE': ['10FrenchGuianahuman'],\n",
       " 'BEF': ['26Asianfermentation'],\n",
       " 'BEH': ['26Asianfermentation'],\n",
       " 'BEI': ['6Africanbeer'],\n",
       " 'BEK': ['6Africanbeer'],\n",
       " 'BEM': ['6Africanbeer'],\n",
       " 'BEN': ['6Africanbeer'],\n",
       " 'BEP': ['1WineEuropeansubclade1'],\n",
       " 'BEQ': ['1WineEuropean'],\n",
       " 'BER': ['1WineEuropean'],\n",
       " 'BES': ['1WineEuropean'],\n",
       " 'BET': ['1WineEuropean'],\n",
       " 'BEV': ['1WineEuropean'],\n",
       " 'BFB': ['1WineEuropeansubclade1'],\n",
       " 'BFC': ['1WineEuropean'],\n",
       " 'BFD': ['1WineEuropeansubclade1'],\n",
       " 'BFE': ['8Mixedorigin'],\n",
       " 'BFG': ['8Mixedorigin'],\n",
       " 'BFH': ['5Frenchdairy'],\n",
       " 'BFK': ['1WineEuropean'],\n",
       " 'BFM': ['8Mixedorigin'],\n",
       " 'BFN': ['5Frenchdairy'],\n",
       " 'BFP': ['4Mediterraneanoak'],\n",
       " 'BFQ': ['1WineEuropean'],\n",
       " 'BFR': ['1WineEuropean'],\n",
       " 'BFS': ['5Frenchdairy'],\n",
       " 'BFT': ['5Frenchdairy'],\n",
       " 'BFV': ['5Frenchdairy'],\n",
       " 'BGA': ['5Frenchdairy'],\n",
       " 'BGB': ['5Frenchdairy'],\n",
       " 'BGC': ['5Frenchdairy'],\n",
       " 'BGD': ['5Frenchdairy'],\n",
       " 'BGE': ['5Frenchdairy'],\n",
       " 'BGF': ['5Frenchdairy'],\n",
       " 'BGG': ['5Frenchdairy'],\n",
       " 'BGH': ['5Frenchdairy'],\n",
       " 'BGI': ['5Frenchdairy'],\n",
       " 'BGK': ['5Frenchdairy'],\n",
       " 'BGL': ['5Frenchdairy'],\n",
       " 'BGM': ['5Frenchdairy'],\n",
       " 'BGN': ['5Frenchdairy'],\n",
       " 'BGP': ['5Frenchdairy'],\n",
       " 'BGQ': ['5Frenchdairy'],\n",
       " 'BGR': ['5Frenchdairy'],\n",
       " 'BGS': ['5Frenchdairy'],\n",
       " 'BGT': ['5Frenchdairy'],\n",
       " 'BGV': ['5Frenchdairy'],\n",
       " 'BHA': ['6Africanbeer'],\n",
       " 'BHB': ['1WineEuropean'],\n",
       " 'BHC': ['1WineEuropean'],\n",
       " 'BHD': ['M1Mosaicregion1'],\n",
       " 'BHE': ['M1Mosaicregion1'],\n",
       " 'BHF': ['1WineEuropean'],\n",
       " 'BHG': ['M2Mosaicregion2'],\n",
       " 'BHH': ['M2Mosaicregion2'],\n",
       " 'BHI': ['M2Mosaicregion2'],\n",
       " 'BHK': ['M1Mosaicregion1'],\n",
       " 'BHL': ['M1Mosaicregion1'],\n",
       " 'BHM': ['M1Mosaicregion1'],\n",
       " 'BHN': ['1WineEuropean'],\n",
       " 'BHP': ['26Asianfermentation'],\n",
       " 'BHQ': ['26Asianfermentation'],\n",
       " 'BHR': ['26Asianfermentation'],\n",
       " 'BHS': ['26Asianfermentation'],\n",
       " 'BHT': ['M1Mosaicregion1'],\n",
       " 'BHV': ['1WineEuropean'],\n",
       " 'BIA': ['1WineEuropean'],\n",
       " 'BIB': ['1WineEuropean'],\n",
       " 'BIC': ['1WineEuropean'],\n",
       " 'BID': ['1WineEuropean'],\n",
       " 'BIE': ['M3Mosaicregion3'],\n",
       " 'BIF': ['1WineEuropean'],\n",
       " 'BIG': ['1WineEuropean'],\n",
       " 'BIH': ['1WineEuropean'],\n",
       " 'BII': ['1WineEuropean'],\n",
       " 'BIK': ['1WineEuropean'],\n",
       " 'BIL': ['1WineEuropean'],\n",
       " 'BIM': ['1WineEuropean'],\n",
       " 'BIN': ['1WineEuropean'],\n",
       " 'BIP': ['1WineEuropean'],\n",
       " 'BIQ': ['1WineEuropean'],\n",
       " 'BIR': ['1WineEuropean'],\n",
       " 'BIS': ['1WineEuropean'],\n",
       " 'BIT': ['1WineEuropean'],\n",
       " 'BKA': ['1WineEuropean'],\n",
       " 'BKC': ['1WineEuropean'],\n",
       " 'BKD': ['1WineEuropean'],\n",
       " 'BKE': ['1WineEuropean'],\n",
       " 'BKF': ['1WineEuropean'],\n",
       " 'BKG': ['1WineEuropean'],\n",
       " 'BKH': ['1WineEuropean'],\n",
       " 'BKI': ['1WineEuropean'],\n",
       " 'BKK': ['1WineEuropean'],\n",
       " 'BKL': ['1WineEuropean'],\n",
       " 'BKM': ['1WineEuropean'],\n",
       " 'BKN': ['1WineEuropean'],\n",
       " 'BKP': ['1WineEuropean'],\n",
       " 'BKQ': ['1WineEuropean'],\n",
       " 'BKR': ['1WineEuropean'],\n",
       " 'BKS': ['1WineEuropean'],\n",
       " 'BKT': ['1WineEuropean'],\n",
       " 'BLA': ['1WineEuropean'],\n",
       " 'BLB': ['1WineEuropean'],\n",
       " 'BLC': ['1WineEuropean'],\n",
       " 'BLD': ['1WineEuropean'],\n",
       " 'BLE': ['1WineEuropean'],\n",
       " 'BLF': ['1WineEuropean'],\n",
       " 'BLG': ['1WineEuropean'],\n",
       " 'BLH': ['1WineEuropean'],\n",
       " 'BLI': ['1WineEuropean'],\n",
       " 'BLK': ['1WineEuropean'],\n",
       " 'BLL': ['1WineEuropean'],\n",
       " 'BLM': ['1WineEuropean'],\n",
       " 'BLN': ['1WineEuropean'],\n",
       " 'BLP': ['1WineEuropean'],\n",
       " 'BLQ': ['1WineEuropean'],\n",
       " 'BLR': ['1WineEuropean'],\n",
       " 'BLS': ['1WineEuropean'],\n",
       " 'BLT': ['1WineEuropean'],\n",
       " 'BLV': ['1WineEuropean'],\n",
       " 'BMA': ['19Malaysian'],\n",
       " 'BMB': ['19Malaysian'],\n",
       " 'BMC': ['19Malaysian'],\n",
       " 'BMD': ['M3Mosaicregion3'],\n",
       " 'BME': ['5Frenchdairy'],\n",
       " 'BMF': ['M3Mosaicregion3'],\n",
       " 'BMG': ['8Mixedorigin'],\n",
       " 'BMH': ['1WineEuropeansubclade2'],\n",
       " 'BMI': ['1WineEuropeansubclade2'],\n",
       " 'BMK': ['1WineEuropeansubclade2'],\n",
       " 'BML': ['8Mixedorigin'],\n",
       " 'BMM': ['8Mixedorigin'],\n",
       " 'BMN': ['1WineEuropeansubclade3'],\n",
       " 'BMQ': ['M3Mosaicregion3'],\n",
       " 'BMR': ['10FrenchGuianahuman'],\n",
       " 'BMV': ['10FrenchGuianahuman'],\n",
       " 'BNA': ['10FrenchGuianahuman'],\n",
       " 'BNB': ['10FrenchGuianahuman'],\n",
       " 'BNC': ['10FrenchGuianahuman'],\n",
       " 'BND': ['10FrenchGuianahuman'],\n",
       " 'BNE': ['1WineEuropean'],\n",
       " 'BNF': ['1WineEuropean'],\n",
       " 'BNG': ['1WineEuropean'],\n",
       " 'BNH': ['1WineEuropean'],\n",
       " 'BNI': ['1WineEuropean'],\n",
       " 'BNK': ['1WineEuropean'],\n",
       " 'BNL': ['1WineEuropean'],\n",
       " 'BNM': ['1WineEuropean'],\n",
       " 'BNN': ['1WineEuropean'],\n",
       " 'BNP': ['1WineEuropean'],\n",
       " 'BNQ': ['1WineEuropean'],\n",
       " 'BNR': ['1WineEuropean'],\n",
       " 'BNT': ['1WineEuropean'],\n",
       " 'BNV': ['1WineEuropean'],\n",
       " 'BPA': ['M3Mosaicregion3'],\n",
       " 'BPB': ['1WineEuropean'],\n",
       " 'BPC': ['1WineEuropean'],\n",
       " 'BPF': ['1WineEuropean'],\n",
       " 'BPG': ['1WineEuropean'],\n",
       " 'BPH': ['1WineEuropean'],\n",
       " 'BPK': ['1WineEuropean'],\n",
       " 'BPL': ['1WineEuropeansubclade4'],\n",
       " 'BPM': ['1WineEuropean'],\n",
       " 'BPP': ['1WineEuropean'],\n",
       " 'BPQ': ['1WineEuropeansubclade1'],\n",
       " 'BPR': ['1WineEuropeansubclade4'],\n",
       " 'BPT': ['1WineEuropean'],\n",
       " 'BPV': ['1WineEuropean'],\n",
       " 'BQA': ['1WineEuropeansubclade1'],\n",
       " 'BQB': ['1WineEuropean'],\n",
       " 'BQC': ['1WineEuropean'],\n",
       " 'BQE': ['1WineEuropean'],\n",
       " 'BQG': ['1WineEuropean'],\n",
       " 'BQH': ['1WineEuropean'],\n",
       " 'BQI': ['1WineEuropean'],\n",
       " 'BQM': ['1WineEuropean'],\n",
       " 'BQN': ['1WineEuropeansubclade1'],\n",
       " 'BQP': ['1WineEuropean'],\n",
       " 'BQQ': ['1WineEuropean'],\n",
       " 'BQR': ['1WineEuropean'],\n",
       " 'BQS': ['1WineEuropean'],\n",
       " 'BQT': ['1WineEuropean'],\n",
       " 'BRA': ['1WineEuropean'],\n",
       " 'BRB': ['1WineEuropean'],\n",
       " 'BRC': ['1WineEuropeansubclade1'],\n",
       " 'BRD': ['1WineEuropean'],\n",
       " 'BRE': ['1WineEuropean'],\n",
       " 'BRF': ['1WineEuropean'],\n",
       " 'BRG': ['26Asianfermentation'],\n",
       " 'BRH': ['1WineEuropean'],\n",
       " 'BRK': ['2Alpechin'],\n",
       " 'BRL': ['M3Mosaicregion3'],\n",
       " 'BRM': ['7Mosaicbeer'],\n",
       " 'BRP': ['11Alebeer'],\n",
       " 'BRQ': ['11Alebeer'],\n",
       " 'BRV': ['M3Mosaicregion3'],\n",
       " 'BSA': ['1WineEuropean'],\n",
       " 'BSB': ['1WineEuropean'],\n",
       " 'BSC': ['1WineEuropean'],\n",
       " 'BSD': ['1WineEuropean'],\n",
       " 'BSE': ['1WineEuropean'],\n",
       " 'BSG': ['1WineEuropean'],\n",
       " 'BSI': ['11Alebeer'],\n",
       " 'BSK': ['M3Mosaicregion3'],\n",
       " 'BSL': ['1WineEuropean'],\n",
       " 'BSM': ['1WineEuropean'],\n",
       " 'BSN': ['1WineEuropean'],\n",
       " 'BSP': ['1WineEuropean'],\n",
       " 'BSR': ['1WineEuropean'],\n",
       " 'BSS': ['1WineEuropean'],\n",
       " 'BST': ['1WineEuropean'],\n",
       " 'BTA': ['M3Mosaicregion3'],\n",
       " 'BTB': ['M3Mosaicregion3'],\n",
       " 'BTC': ['8Mixedorigin'],\n",
       " 'BTD': ['8Mixedorigin'],\n",
       " 'BTE': ['8Mixedorigin'],\n",
       " 'BTF': ['8Mixedorigin'],\n",
       " 'BTG': ['4Mediterraneanoak'],\n",
       " 'BTH': ['1WineEuropean'],\n",
       " 'BTI': ['M3Mosaicregion3'],\n",
       " 'BTL': ['1WineEuropean'],\n",
       " 'BTN': ['1WineEuropean'],\n",
       " 'BTP': ['1WineEuropean'],\n",
       " 'BTR': ['1WineEuropean'],\n",
       " 'BTS': ['25Sake'],\n",
       " 'BTT': ['3Brazilianbioethanol'],\n",
       " 'BTV': ['3Brazilianbioethanol'],\n",
       " 'BVA': ['3Brazilianbioethanol'],\n",
       " 'BVB': ['3Brazilianbioethanol'],\n",
       " 'BVC': ['3Brazilianbioethanol'],\n",
       " 'BVD': ['3Brazilianbioethanol'],\n",
       " 'BVE': ['3Brazilianbioethanol'],\n",
       " 'BVF': ['3Brazilianbioethanol'],\n",
       " 'BVG': ['3Brazilianbioethanol'],\n",
       " 'BVH': ['3Brazilianbioethanol'],\n",
       " 'BVK': ['1WineEuropean'],\n",
       " 'BVM': ['1WineEuropean'],\n",
       " 'BVT': ['1WineEuropeansubclade4'],\n",
       " 'BVV': ['1WineEuropeansubclade4'],\n",
       " 'CAA': ['1WineEuropeansubclade4'],\n",
       " 'CAB': ['1WineEuropeansubclade4'],\n",
       " 'CAC': ['1WineEuropeansubclade4'],\n",
       " 'CAD': ['1WineEuropeansubclade4'],\n",
       " 'CAE': ['1WineEuropean'],\n",
       " 'CAF': ['1WineEuropean'],\n",
       " 'CAH': ['1WineEuropean'],\n",
       " 'CAI': ['1WineEuropean'],\n",
       " 'CAK': ['1WineEuropean'],\n",
       " 'CAL': ['1WineEuropean'],\n",
       " 'CAM': ['1WineEuropeansubclade4'],\n",
       " 'CAN': ['1WineEuropeansubclade4'],\n",
       " 'CAP': ['1WineEuropeansubclade4'],\n",
       " 'CAQ': ['1WineEuropeansubclade4'],\n",
       " 'CAR': ['1WineEuropeansubclade4'],\n",
       " 'CAS': ['1WineEuropeansubclade4'],\n",
       " 'CAT': ['1WineEuropeansubclade4'],\n",
       " 'CAV': ['1WineEuropeansubclade4'],\n",
       " 'CBA': ['1WineEuropean'],\n",
       " 'CBB': ['1WineEuropeansubclade4'],\n",
       " 'CBC': ['1WineEuropeansubclade4'],\n",
       " 'CBD': ['1WineEuropeansubclade4'],\n",
       " 'CBE': ['1WineEuropeansubclade4'],\n",
       " 'CBF': ['8Mixedorigin'],\n",
       " 'CBG': ['1WineEuropean'],\n",
       " 'CBH': ['1WineEuropean'],\n",
       " 'CBI': ['1WineEuropeansubclade1'],\n",
       " 'CBK': ['M3Mosaicregion3'],\n",
       " 'CBL': ['M3Mosaicregion3'],\n",
       " 'CBM': ['M3Mosaicregion3'],\n",
       " 'CBN': ['7Mosaicbeer'],\n",
       " 'CBR': ['11Alebeer'],\n",
       " 'CCC': ['21Ecuadorean'],\n",
       " 'CCD': ['21Ecuadorean'],\n",
       " 'CCE': ['21Ecuadorean'],\n",
       " 'CCF': ['21Ecuadorean'],\n",
       " 'CCG': ['21Ecuadorean'],\n",
       " 'CCH': ['21Ecuadorean'],\n",
       " 'CCI': ['21Ecuadorean'],\n",
       " 'CCL': ['4Mediterraneanoak'],\n",
       " 'CCM': ['1WineEuropeansubclade1'],\n",
       " 'CCN': ['18FarEastAsia'],\n",
       " 'CCQ': ['18FarEastAsia'],\n",
       " 'CCR': ['18FarEastAsia'],\n",
       " 'CCS': ['18FarEastAsia'],\n",
       " 'CCT': ['24Asianislands'],\n",
       " 'CCV': ['24Asianislands'],\n",
       " 'CDA': ['24Asianislands'],\n",
       " 'CDC': ['M3Mosaicregion3'],\n",
       " 'CDD': ['M3Mosaicregion3'],\n",
       " 'CDE': ['M3Mosaicregion3'],\n",
       " 'CDF': ['22FarEastRussian'],\n",
       " 'CDG': ['22FarEastRussian'],\n",
       " 'CDH': ['22FarEastRussian'],\n",
       " 'CDI': ['22FarEastRussian'],\n",
       " 'CDK': ['M3Mosaicregion3'],\n",
       " 'CDL': ['18FarEastAsia'],\n",
       " 'CDM': ['M2Mosaicregion2'],\n",
       " 'CDR': ['8Mixedorigin'],\n",
       " 'CDS': ['1WineEuropean'],\n",
       " 'CDT': ['2Alpechin'],\n",
       " 'CDV': ['1WineEuropean'],\n",
       " 'CEA': ['1WineEuropean'],\n",
       " 'CEB': ['1WineEuropean'],\n",
       " 'CEC': ['1WineEuropean'],\n",
       " 'CED': ['1WineEuropean'],\n",
       " 'CEE': ['1WineEuropean'],\n",
       " 'CEF': ['1WineEuropean'],\n",
       " 'CEG': ['17Taiwanese'],\n",
       " 'CEI': ['17Taiwanese'],\n",
       " 'CEK': ['26Asianfermentation'],\n",
       " 'CEL': ['26Asianfermentation'],\n",
       " 'CEM': ['M3Mosaicregion3'],\n",
       " 'CEN': ['M3Mosaicregion3'],\n",
       " 'CEQ': ['M3Mosaicregion3'],\n",
       " 'CER': ['6Africanbeer'],\n",
       " 'CES': ['6Africanbeer'],\n",
       " 'CET': ['6Africanbeer'],\n",
       " 'CEV': ['26Asianfermentation'],\n",
       " 'CFA': ['13Africanpalmwine'],\n",
       " 'CFB': ['1WineEuropean'],\n",
       " 'CFC': ['11Alebeer'],\n",
       " 'CFD': ['11Alebeer'],\n",
       " 'CFE': ['1WineEuropean'],\n",
       " 'CFF': ['7Mosaicbeer'],\n",
       " 'CFG': ['11Alebeer'],\n",
       " 'CFH': ['11Alebeer'],\n",
       " 'CFI': ['8Mixedorigin'],\n",
       " 'CFM': ['11Alebeer'],\n",
       " 'CFN': ['8Mixedorigin'],\n",
       " 'CFP': ['8Mixedorigin'],\n",
       " 'CFQ': ['1WineEuropean'],\n",
       " 'CFR': ['1WineEuropean'],\n",
       " 'CFS': ['1WineEuropean'],\n",
       " 'CFT': ['M2Mosaicregion2'],\n",
       " 'CFV': ['1WineEuropean'],\n",
       " 'CGB': ['M3Mosaicregion3'],\n",
       " 'CGC': ['11Alebeer'],\n",
       " 'CGD': ['2Alpechin'],\n",
       " 'CGE': ['2Alpechin'],\n",
       " 'CGG': ['26Asianfermentation'],\n",
       " 'CGI': ['26Asianfermentation'],\n",
       " 'CGK': ['26Asianfermentation'],\n",
       " 'CGL': ['26Asianfermentation'],\n",
       " 'CGN': ['26Asianfermentation'],\n",
       " 'CGP': ['26Asianfermentation'],\n",
       " 'CGQ': ['26Asianfermentation'],\n",
       " 'CGR': ['26Asianfermentation'],\n",
       " 'CGS': ['1WineEuropean'],\n",
       " 'CGT': ['1WineEuropeansubclade1'],\n",
       " 'CGV': ['1WineEuropean'],\n",
       " 'CHA': ['1WineEuropean'],\n",
       " 'CHB': ['1WineEuropean'],\n",
       " 'CHC': ['1WineEuropean'],\n",
       " 'CHD': ['1WineEuropean'],\n",
       " 'CHE': ['1WineEuropean'],\n",
       " 'CHF': ['1WineEuropean'],\n",
       " 'CHG': ['8Mixedorigin'],\n",
       " 'CHH': ['8Mixedorigin'],\n",
       " 'CHI': ['8Mixedorigin'],\n",
       " 'CHK': ['8Mixedorigin'],\n",
       " 'CHL': ['8Mixedorigin'],\n",
       " 'CHM': ['26Asianfermentation'],\n",
       " 'CHN': ['M3Mosaicregion3'],\n",
       " 'CHP': ['M3Mosaicregion3'],\n",
       " 'CHQ': ['M3Mosaicregion3'],\n",
       " 'CHR': ['8Mixedorigin'],\n",
       " 'CHS': ['M3Mosaicregion3'],\n",
       " 'CHT': ['M3Mosaicregion3'],\n",
       " 'CHV': ['M1Mosaicregion1'],\n",
       " 'CIA': ['1WineEuropean'],\n",
       " 'CIB': ['1WineEuropean'],\n",
       " 'CIC': ['M3Mosaicregion3'],\n",
       " 'CID': ['M3Mosaicregion3'],\n",
       " 'CIE': ['8Mixedorigin'],\n",
       " 'CIF': ['M3Mosaicregion3'],\n",
       " 'CIG': ['M3Mosaicregion3'],\n",
       " 'CIH': ['M3Mosaicregion3'],\n",
       " 'CII': ['M3Mosaicregion3'],\n",
       " 'CIK': ['M3Mosaicregion3'],\n",
       " 'CIL': ['M3Mosaicregion3'],\n",
       " 'CIM': ['1WineEuropeansubclade3'],\n",
       " 'CIN': ['1WineEuropeansubclade3'],\n",
       " 'CIP': ['1WineEuropeansubclade3'],\n",
       " 'CIQ': ['1WineEuropeansubclade3'],\n",
       " 'CIR': ['1WineEuropeansubclade3'],\n",
       " 'CIS': ['1WineEuropeansubclade3'],\n",
       " 'CIT': ['1WineEuropeansubclade3'],\n",
       " 'CIV': ['1WineEuropeansubclade3'],\n",
       " 'CKA': ['M3Mosaicregion3'],\n",
       " 'CKB': ['1WineEuropeansubclade3'],\n",
       " 'CKC': ['1WineEuropeansubclade3'],\n",
       " 'CKD': ['1WineEuropeansubclade3'],\n",
       " 'CKE': ['1WineEuropeansubclade3'],\n",
       " 'CKF': ['1WineEuropeansubclade3'],\n",
       " 'CKG': ['1WineEuropeansubclade3'],\n",
       " 'CKH': ['1WineEuropeansubclade3'],\n",
       " 'CKI': ['1WineEuropeansubclade3'],\n",
       " 'CKK': ['8Mixedorigin'],\n",
       " 'CKL': ['1WineEuropeansubclade3'],\n",
       " 'CKM': ['1WineEuropeansubclade3'],\n",
       " 'CKN': ['M3Mosaicregion3'],\n",
       " 'CKP': ['1WineEuropeansubclade3'],\n",
       " 'CKQ': ['1WineEuropeansubclade3'],\n",
       " 'CKS': ['8Mixedorigin'],\n",
       " 'CKT': ['1WineEuropeansubclade3'],\n",
       " 'CKV': ['5Frenchdairy'],\n",
       " 'CLA': ['8Mixedorigin'],\n",
       " 'CLB': ['4Mediterraneanoak'],\n",
       " 'CLC': ['4Mediterraneanoak'],\n",
       " 'CLD': ['4Mediterraneanoak'],\n",
       " 'CLG': ['M3Mosaicregion3'],\n",
       " 'CLH': ['1WineEuropean'],\n",
       " 'CLI': ['25Sake'],\n",
       " 'CLK': ['25Sake'],\n",
       " 'CLL': ['25Sake'],\n",
       " 'CLM': ['25Sake'],\n",
       " 'CLN': ['25Sake'],\n",
       " 'CLP': ['25Sake'],\n",
       " 'CLQ': ['25Sake'],\n",
       " 'CLR': ['25Sake'],\n",
       " 'CLS': ['25Sake'],\n",
       " 'CLT': ['25Sake'],\n",
       " 'CLV': ['25Sake'],\n",
       " 'CMB': ['25Sake'],\n",
       " 'CMC': ['25Sake'],\n",
       " 'CMD': ['25Sake'],\n",
       " 'CME': ['25Sake'],\n",
       " 'CMF': ['25Sake'],\n",
       " 'CMG': ['M3Mosaicregion3'],\n",
       " 'CMH': ['25Sake'],\n",
       " 'CMI': ['25Sake'],\n",
       " 'CMK': ['25Sake'],\n",
       " 'CML': ['25Sake'],\n",
       " 'CMM': ['25Sake'],\n",
       " 'CMN': ['25Sake'],\n",
       " 'CMP': ['21Ecuadorean'],\n",
       " 'CMQ': ['21Ecuadorean'],\n",
       " 'CMT': ['3Brazilianbioethanol'],\n",
       " 'CNB': ['3Brazilianbioethanol'],\n",
       " 'CNC': ['M3Mosaicregion3'],\n",
       " 'CNE': ['3Brazilianbioethanol'],\n",
       " 'CNF': ['3Brazilianbioethanol'],\n",
       " 'CNG': ['3Brazilianbioethanol'],\n",
       " 'CNH': ['3Brazilianbioethanol'],\n",
       " 'CNI': ['3Brazilianbioethanol'],\n",
       " 'CNK': ['3Brazilianbioethanol'],\n",
       " 'CNL': ['3Brazilianbioethanol'],\n",
       " 'CNM': ['3Brazilianbioethanol'],\n",
       " 'CNN': ['3Brazilianbioethanol'],\n",
       " 'CNP': ['3Brazilianbioethanol'],\n",
       " 'CNQ': ['3Brazilianbioethanol'],\n",
       " 'CNR': ['3Brazilianbioethanol'],\n",
       " 'CNS': ['3Brazilianbioethanol'],\n",
       " 'CNT': ['3Brazilianbioethanol'],\n",
       " 'CNV': ['3Brazilianbioethanol'],\n",
       " 'CPB': ['M1Mosaicregion1'],\n",
       " 'CPC': ['M3Mosaicregion3'],\n",
       " 'CPD': ['7Mosaicbeer'],\n",
       " 'CPE': ['1WineEuropean'],\n",
       " 'CPF': ['1WineEuropean'],\n",
       " 'CPG': ['2Alpechin'],\n",
       " 'CPI': ['9Mexicanagave'],\n",
       " 'CPK': ['9Mexicanagave'],\n",
       " 'CPL': ['9Mexicanagave'],\n",
       " 'CPM': ['9Mexicanagave'],\n",
       " 'CPN': ['9Mexicanagave'],\n",
       " 'CPQ': ['9Mexicanagave'],\n",
       " 'CPR': ['9Mexicanagave'],\n",
       " 'CPS': ['13Africanpalmwine'],\n",
       " 'CPT': ['13Africanpalmwine'],\n",
       " 'CQA': ['13Africanpalmwine'],\n",
       " 'CQB': ['13Africanpalmwine'],\n",
       " 'CQC': ['12WestAfricancocoa'],\n",
       " 'CQD': ['12WestAfricancocoa'],\n",
       " 'CQE': ['12WestAfricancocoa'],\n",
       " 'CQF': ['12WestAfricancocoa'],\n",
       " 'CQG': ['12WestAfricancocoa'],\n",
       " 'CQH': ['12WestAfricancocoa'],\n",
       " 'CQI': ['12WestAfricancocoa'],\n",
       " 'CQK': ['12WestAfricancocoa'],\n",
       " 'CQL': ['12WestAfricancocoa'],\n",
       " 'CQM': ['12WestAfricancocoa'],\n",
       " 'CQN': ['12WestAfricancocoa'],\n",
       " 'CQP': ['12WestAfricancocoa'],\n",
       " 'CQQ': ['25Sake'],\n",
       " 'CQR': ['25Sake'],\n",
       " 'CQT': ['1WineEuropean'],\n",
       " 'CQV': ['1WineEuropean'],\n",
       " 'CRA': ['1WineEuropean'],\n",
       " 'CRB': ['1WineEuropean'],\n",
       " 'CRC': ['1WineEuropean'],\n",
       " 'CRD': ['1WineEuropean'],\n",
       " 'CRE': ['M3Mosaicregion3'],\n",
       " 'CRG': ['1WineEuropean'],\n",
       " 'CRH': ['1WineEuropean'],\n",
       " 'CRI': ['1WineEuropean'],\n",
       " 'CRK': ['1WineEuropean'],\n",
       " 'CRL': ['1WineEuropean'],\n",
       " 'SACE_GAL': ['13Africanpalmwine'],\n",
       " 'SACE_GAP': ['1WineEuropean'],\n",
       " 'SACE_GAV': ['M3Mosaicregion3'],\n",
       " 'SACE_MAA': ['M2Mosaicregion2'],\n",
       " 'SACE_MAB': ['M3Mosaicregion3'],\n",
       " 'SACE_MAL': ['19Malaysian'],\n",
       " 'SACE_YAA': ['1WineEuropean'],\n",
       " 'SACE_YAB': ['7Mosaicbeer'],\n",
       " 'SACE_YAC': ['13Africanpalmwine'],\n",
       " 'SACE_YAD': ['1WineEuropean'],\n",
       " 'SACE_YAE': ['2Alpechin'],\n",
       " 'SACE_YAF': ['1WineEuropeansubclade4'],\n",
       " 'SACE_YAG': ['7Mosaicbeer'],\n",
       " 'SACE_YAK': ['M3Mosaicregion3'],\n",
       " 'SACE_YAL': ['M3Mosaicregion3'],\n",
       " 'SACE_YAM': ['1WineEuropean'],\n",
       " 'SACE_YAN': ['M3Mosaicregion3'],\n",
       " 'SACE_YAP': ['M3Mosaicregion3'],\n",
       " 'SACE_YAQ': ['M3Mosaicregion3'],\n",
       " 'SACE_YAR': ['M3Mosaicregion3'],\n",
       " 'SACE_YAT': ['M3Mosaicregion3'],\n",
       " 'SACE_YAU': ['M3Mosaicregion3'],\n",
       " 'SACE_YAV': ['M3Mosaicregion3'],\n",
       " 'SACE_YAW': ['M3Mosaicregion3'],\n",
       " 'SACE_YAY': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YAZ': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBA': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBB': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBE': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBF': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBG': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBH': ['1WineEuropeansubclade2'],\n",
       " 'SACE_YBJ': ['M3Mosaicregion3'],\n",
       " 'SACE_YBK': ['M3Mosaicregion3'],\n",
       " 'SACE_YBL': ['M3Mosaicregion3'],\n",
       " 'SACE_YBM': ['M3Mosaicregion3'],\n",
       " 'SACE_YBN': ['1WineEuropean'],\n",
       " 'SACE_YBO': ['1WineEuropean'],\n",
       " 'SACE_YBP': ['13Africanpalmwine'],\n",
       " 'SACE_YBR': ['2Alpechin'],\n",
       " 'SACE_YBS': ['23NorthAmericanoak'],\n",
       " 'SACE_YBT': ['M3Mosaicregion3'],\n",
       " 'SACE_YBU': ['M3Mosaicregion3'],\n",
       " 'SACE_YBX': ['1WineEuropean'],\n",
       " 'SACE_YBY': ['1WineEuropean'],\n",
       " 'SACE_YBZ': ['M3Mosaicregion3'],\n",
       " 'SACE_YCA': ['1WineEuropean'],\n",
       " 'SACE_YCC': ['M2Mosaicregion2'],\n",
       " 'SACE_YCD': ['1WineEuropean'],\n",
       " 'SACE_YCE': ['M3Mosaicregion3'],\n",
       " 'SACE_YCF': ['M2Mosaicregion2'],\n",
       " 'SACE_YCH': ['M3Mosaicregion3'],\n",
       " 'SACE_YCI': ['M1Mosaicregion1'],\n",
       " 'SACE_YCJ': ['26Asianfermentation'],\n",
       " 'SACE_YCK': ['26Asianfermentation'],\n",
       " 'SACE_YCM': ['24Asianislands'],\n",
       " 'SACE_YCP': ['1WineEuropean'],\n",
       " 'SACE_YCQ': ['1WineEuropean'],\n",
       " 'SACE_YCR': ['18FarEastAsia'],\n",
       " 'SACE_YCS': ['M3Mosaicregion3'],\n",
       " 'SACE_YCT': ['M2Mosaicregion2'],\n",
       " 'SACE_YCU': ['23NorthAmericanoak'],\n",
       " 'SACE_YCV': ['13Africanpalmwine'],\n",
       " 'SACE_YCY': ['19Malaysian'],\n",
       " 'SACE_YCZ': ['M2Mosaicregion2'],\n",
       " 'SACE_YDA': ['26Asianfermentation'],\n",
       " 'SACE_YDC': ['1WineEuropean'],\n",
       " 'SACE_YDD': ['7Mosaicbeer'],\n",
       " 'SACE_YDE': ['24Asianislands'],\n",
       " 'SACE_YDF': ['1WineEuropean'],\n",
       " 'SACE_YDG': ['M3Mosaicregion3'],\n",
       " 'SACE_YDH': ['M3Mosaicregion3'],\n",
       " 'SACE_YDJ': ['1WineEuropean'],\n",
       " 'SACE_YDK': ['26Asianfermentation'],\n",
       " 'SACE_YDL': ['M3Mosaicregion3'],\n",
       " 'SACE_YDM': ['2Alpechin'],\n",
       " 'SACE_YDN': ['M3Mosaicregion3'],\n",
       " 'SACE_YDO': ['1WineEuropean']}"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#group Scer strains by population\n",
    "\n",
    "pops = pd.read_csv('/Users/clairedubin/sacc/external_datasets/1011_pops_tableS1.csv', skiprows=3)\n",
    "pops = pops[['Standardized name', 'Clades']].dropna()\n",
    "pops['clade'] = pops['Clades'].str.replace('[^\\w]','')\n",
    "\n",
    "pop_dict = pops.groupby('clade')['Standardized name'].apply(list).to_dict()\n",
    "pop_dict\n",
    "\n",
    "gene_dict = pops.groupby('Standardized name')['clade'].apply(list).to_dict()\n",
    "gene_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "#make Scer files by population\n",
    "scer_outpath = '/Users/clairedubin/sacc/carly_genes/alignments/1011pops_unaligned/'\n",
    "all_genes = '/Users/clairedubin/sacc/external_datasets/allReferenceGenesWithSNPsAndIndelsInferred/'\n",
    "\n",
    "for clade in pop_list:\n",
    "    \n",
    "    if not os.path.exists(scer_outpath+clade):\n",
    "        os.mkdir(scer_outpath+clade)\n",
    "        \n",
    "for file in os.listdir(all_genes):\n",
    "    \n",
    "    gene = file.strip('.fasta')\n",
    "    for record in SeqIO.parse(all_genes+file, 'fasta'):\n",
    "        strain = record.description.split('_'+gene)[0]\n",
    "        \n",
    "        if strain not in gene_dict:\n",
    "            continue\n",
    "        \n",
    "        clade = gene_dict[strain][0]\n",
    "        \n",
    "        f = open(scer_outpath+clade+'/'+gene+'.fa', 'a')\n",
    "        f.write('>'+record.description+'\\n'+str(record.seq)+'\\n')\n",
    "        f.close()\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "to_remove = []\n",
    "\n",
    "for clade in os.listdir(scer_outpath):\n",
    "    popsize = len(pop_dict[clade])\n",
    "    \n",
    "    for file in os.listdir(scer_outpath+clade):        \n",
    "        if len([r for r in SeqIO.parse(scer_outpath+clade+'/'+file, 'fasta')]) < 0.75*popsize:\n",
    "            to_remove += [scer_outpath+clade+'/'+file]\n",
    "\n",
    "to_remove"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### S. paradoxus European population - Bergstrom, et al. 2014\n",
    "\n",
    "Write file for each gene using annotations and assemblies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "#use only European population strains\n",
    "spar_euro_strains = ['Q59.1', \n",
    "                'Z1.1',\n",
    "                'Q95.3',\n",
    "                'S36.7', \n",
    "                'Y9.6', \n",
    "                'W7', \n",
    "                'Y8.5', \n",
    "                'Z1', \n",
    "                'T21.4',\n",
    "                'Y7',\n",
    "                'Y6.5']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_assemblies = '/Users/clairedubin/sacc/external_datasets/SGRP2-assemblies_Jun25/'\n",
    "all_annotations = '/Users/clairedubin/sacc/external_datasets/orf-annotation-gffs-SGRP2_Jun25/'\n",
    "spar_annotartions = '/Users/clairedubin/sacc/external_datasets/Spar.gff.txt'\n",
    "\n",
    "euro_spar_output_dir = '/Users/clairedubin/sacc/carly_genes/alignments/BergstromSpar_ORFs_unaligned/'\n",
    "\n",
    "#map genes to homologs\n",
    "gene_dict = {}\n",
    "spar_annotations = '/Users/clairedubin/sacc/external_datasets/Spar.gff.txt'\n",
    "\n",
    "f = open(spar_annotations, 'r')\n",
    "for line in f.readlines():\n",
    "    gene = line.split('SGD=')[1].split(';')[0]\n",
    "    homolog = line.split('Gene=')[1].split(';')[0]\n",
    "    gene_dict[homolog] = gene\n",
    "f.close()    \n",
    "    \n",
    "for file in os.listdir(all_annotations):\n",
    "    \n",
    "    if '_Sc_' in file:\n",
    "        continue\n",
    "    \n",
    "    suffix = file.split('orf-annotation_')[1][:-4]\n",
    "    strain = suffix.split('_Sp_')[1].strip('SGRP2-assembly.gff')\n",
    "    \n",
    "    if strain in spar_euro_strains:\n",
    "        f = open(all_annotations+file, 'r')\n",
    "        annotation_dict = {}\n",
    "        for i, line in enumerate(f.readlines()):  \n",
    "            homolog = line.split('homolog=')[1].strip('\\n')\n",
    "            if homolog in gene_dict and gene_dict[homolog]:\n",
    "                gene = gene_dict[homolog]\n",
    "                split = line.split('\\t')\n",
    "                scaff, start, stop, direction = split[0], split[3], split[4], split[6]\n",
    "                annotation_dict[gene] = [scaff, start, stop, direction]   \n",
    "        f.close()\n",
    "                \n",
    "        f=Fasta(all_assemblies+suffix+'.fa')   \n",
    "        for gene in annotation_dict:            \n",
    "            chrom, start, stop, direction = annotation_dict[gene]\n",
    "            seq = f.sequence({'chr': chrom, 'start': int(start), 'stop': int(stop), 'strand': direction})\n",
    "            to_write = '>' + gene + '_Sp_' + strain + '\\n' + seq + '\\n'\n",
    "\n",
    "            if len(seq) > 50:\n",
    "                newfile=open(euro_spar_output_dir+gene+'.fa', 'a')\n",
    "                newfile.write(to_write)\n",
    "                newfile.close()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### S. paradoxus North American subpopulation B - Durand, et al. 2011\n",
    "\n",
    "Write file for each gene based on assemblies and annotations for strains in subpopulation B (largest subpop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'SpA': ['A01', 'B01', 'C01', 'B02', 'D01'],\n",
       " 'SpB': ['B06', 'D06', 'C05', 'C04', 'D04', 'D05', 'B05', 'A05', 'C06', 'A06'],\n",
       " 'SpC': ['C03', 'A04', 'B03', 'B04', 'A03', 'A02', 'D02', 'D03', 'C02']}"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#parse populations\n",
    "pop_info = '/Users/clairedubin/sacc/external_datasets/durand_spar_NA_raw/infos_lib.txt'\n",
    "\n",
    "NA_spar_pops = {'SpA':[], 'SpB':[], 'SpC': []}\n",
    "\n",
    "f = open(pop_info, 'r')\n",
    "for line in f.readlines():\n",
    "    if 'SpA' in line:\n",
    "        NA_spar_pops['SpA'] += [line.split('\\t')[1].replace('S','')]\n",
    "    elif 'SpB' in line:\n",
    "        NA_spar_pops['SpB'] += [line.split('\\t')[1].replace('S','')]\n",
    "    elif 'SpC' in line:\n",
    "        NA_spar_pops['SpC'] += [line.split('\\t')[1].replace('S','')]\n",
    "        \n",
    "NA_spar_pops"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "NA_spar_output_dir = '/Users/clairedubin/sacc/carly_genes/alignments/DurandSparB_ORFs_unaligned/'\n",
    "all_assemblies = '/Users/clairedubin/sacc/external_datasets/durand_spar_NA_raw/fasta_genomes/'\n",
    "all_annotations = '/Users/clairedubin/sacc/external_datasets/durand_spar_NA_raw/gff_genes/1_gff_genes_coord/'\n",
    "\n",
    "for strain in [i for i in os.listdir(all_annotations) if i.split('_')[0] in NA_spar_pops['SpB']]:\n",
    "\n",
    "    f = open(all_annotations+strain, 'r')\n",
    "    annotation_dict = {}\n",
    "\n",
    "    for i, line in enumerate(f.readlines()):\n",
    "        if 'gene' in line:\n",
    "            gene = line.split('gene_id=')[1].strip('\\n')\n",
    "            if 'unknown' in gene: \n",
    "                continue\n",
    "            split = line.split('\\t')\n",
    "            scaff, start, stop, direction = split[0], split[3], split[4], split[6]\n",
    "            annotation_dict[gene] = [scaff, start, stop, direction]\n",
    "    f.close()\n",
    "\n",
    "    strain = strain.strip('_genes_aug.gff')\n",
    "    f=Fasta(all_assemblies+strain+'_genome_200.fasta')       \n",
    "\n",
    "    for gene in annotation_dict:\n",
    "        chrom, start, stop, direction = annotation_dict[gene]\n",
    "        seq = f.sequence({'chr': chrom, 'start': int(start), 'stop': int(stop), 'strand': direction})\n",
    "\n",
    "        to_write ='>' + gene + '_Sp_' + strain + '\\n' + seq + '\\n'\n",
    "\n",
    "        if len(seq) > 50:\n",
    "            newfile=open(NA_spar_output_dir+gene+'.fa', \"a\")\n",
    "            newfile.write(to_write)\n",
    "            newfile.close()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Merging S. paradoxus European population with each S. cerevisiae population"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "scer_eurospar_outdir = '/Users/clairedubin/sacc/carly_genes/alignments/1011pops_EuroSpar_unaligned/'\n",
    "\n",
    "for clade in pop_dict:\n",
    "        \n",
    "    if not os.path.exists(scer_eurospar_outdir+clade):\n",
    "        os.mkdir(scer_eurospar_outdir+clade)\n",
    "    \n",
    "    for file in os.listdir(scer_outpath+clade):\n",
    "        \n",
    "        if os.path.exists(euro_spar_output_dir+file):\n",
    "                        \n",
    "            sp = open(euro_spar_output_dir+file, 'r')\n",
    "            sp_data = sp.read()\n",
    "            sp.close()\n",
    "            \n",
    "            sc = open(scer_outpath+clade+'/'+file, 'r')\n",
    "            sc_data = sc.read()\n",
    "            sc.close()\n",
    "            \n",
    "            sc_sp = open(scer_eurospar_outdir+clade+'/'+file, 'a')\n",
    "            sc_sp.write(sc_data)\n",
    "            sc_sp.write(sp_data)\n",
    "            sc_sp.close()        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Merging S. paradoxus North American population with each S. cerevisiae population"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "scer_NAspar_outdir = '/Users/clairedubin/sacc/carly_genes/alignments/1011pops_NASpar_unaligned/'\n",
    "\n",
    "for clade in pop_dict:\n",
    "        \n",
    "    if not os.path.exists(scer_NAspar_outdir+clade):\n",
    "        os.mkdir(scer_NAspar_outdir+clade)\n",
    "    \n",
    "    for file in os.listdir(scer_outpath+clade):\n",
    "        \n",
    "        if os.path.exists(NA_spar_output_dir+file):\n",
    "                        \n",
    "            sp = open(NA_spar_output_dir+file, 'r')\n",
    "            sp_data = sp.read()\n",
    "            sp.close()\n",
    "            \n",
    "            sc = open(scer_outpath+clade+'/'+file, 'r')\n",
    "            sc_data = sc.read()\n",
    "            sc.close()\n",
    "            \n",
    "            sc_sp = open(scer_NAspar_outdir+clade+'/'+file, 'a')\n",
    "            sc_sp.write(sc_data)\n",
    "            sc_sp.write(sp_data)\n",
    "            sc_sp.close()   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}