Практикум 8

      import requests, sys

      server = 'https://rest.ensembl.org'
      seq = []
      for coord in start:
        ext = f"/sequence/region/human/1:{coord-6}..{coord+6}:1?expand_3prime=0;expand_5prime=0"
        r = requests.get(server+ext, headers={ "Content-Type" : "text/x-fasta"})

        if not r.ok:
          r.raise_for_status()
          sys.exit()

        seq.append(r.text.split('\n')[1])
      

        import numpy as np

        train = seq[0:100]
        test = seq[100:200]
        
        train_spl = np.array([list(i) for i in train])
        test_spl = np.array([list(i) for i in test])
      

    mask_A = train_spl == 'A'
    count_A = mask_A.sum(axis=0)
    mask_T = train_spl == 'T'
    count_T = mask_T.sum(axis=0)
    mask_G = train_spl == 'G'
    count_G = mask_G.sum(axis=0)
    mask_C = train_spl == 'C'
    count_C = mask_C.sum(axis=0)
    

      gc = 0.423 #GC состав из базы данных NCBI
      eps = 0.1
      
      prob_A = (count_A+eps)/(len(train_spl)+eps)
      prob_T = (count_T+eps)/(len(train_spl)+eps)
      prob_G = (count_G+eps)/(len(train_spl)+eps)
      prob_C = (count_C+eps)/(len(train_spl)+eps)

      A = np.log(2*prob_A/(1-gc))
      T = np.log(2*prob_T/(1-gc))
      G = np.log(2*prob_G/gc)
      C = np.log(2*prob_C/gc)

      pwm = pd.DataFrame([A, T, G, C], ['A', 'T', 'G', 'C'], [i for i in range(1, 14)])
      

      positive = []
      for i in test_spl:
        positive.append(np.sum([pwm.loc[j].iloc[k] for k, j in enumerate(i)]))
      

        import random
        from Bio import SeqIO

        neg_df = pd.read_csv('./sarsatg.tsv', sep = '\t')

        random.seed(422)
        
        clean_negative = neg_df[(neg_df.Strand == '+') & (neg_df.Pattern != 'start-tr')].reset_index(drop = True)
        negative_random = random.choices(clean_negative.index, k = 100)
        
        coords_atg = clean_negative.iloc[negative_random, [0, 1]].reset_index(drop = True)
        coords_atg.Start = coords_atg.Start - 8
        coords_atg.End = coords_atg.End + 3
        
        iterator = SeqIO.parse('./sars.fasta', 'fasta')
        genome = next(iterator)
        
        negative_seqs = []
        for i in coords_atg.index:
            negative_seqs.append(genome.seq[int(coords_atg.iloc[i, 0]):int(coords_atg.iloc[i, 1])])
            
        with open('./negative_seqs.fasta', 'w') as negative_seqs_fasta:
            for i in negative_seqs:
                print(i, file = negative_seqs_fasta)
        
        negative = []
        for i in negative_seqs:
            negative.append(sum([pwm.loc[j].iloc[k] for k, j in enumerate(i)]))
        

          ic_dict = {'A': [], 'T': [], 'G': [], 'C': []}
          proba = {'A': (1-gc)/2, 'T': (1-gc)/2, 'G': gc/2, 'C': gc/2}

          transposed = np.transpose(list(map(lambda x: [i for i in x.__str__()], train)))

          df = pd.DataFrame(transposed)
          for i in df.index:
              values = df.iloc[i, :].value_counts()
              
              for k, v in proba.items():
                  
                  try:
                      ic_dict[k].append(round((values.loc[k]/len(train_spl))*np.log2(values.loc[k]/(len(train_spl)*v)), 3))
                  except KeyError:
                      ic_dict[k].append(0)
                      
          ic = np.transpose(pd.DataFrame(ic_dict, [i for i in range(1, 14)]))