df oktoberfest

data/data_processing.py

@@ -33,45 +33,45 @@ ALPHABET_UNMOD = {
 ALPHABET_UNMOD_REV = {v: k for k, v in ALPHABET_UNMOD.items()}
 
 
-# def align(dataset, reference, column_dataset, column_ref, seq_data, seq_ref):
-#     dataset_ref=dataset[dataset['state']=='train']
-#     dataset_unique = dataset_ref[[seq_data,column_dataset]].groupby(seq_data).mean()
-#     print('unique',len(dataset_unique))
-#     reference_unique = reference[[seq_ref,column_ref]].groupby(seq_ref).mean()
-#     seq_ref = reference_unique.index
-#     seq_common = dataset_unique.index
-#     seq_ref = seq_ref.tolist()
-#     seq_common = seq_common.tolist()
-#
-#     seq_ref = [tuple(l) for l in seq_ref]
-#     seq_common = [tuple(l) for l in seq_common]
-#
-#     ind_dict_ref = dict((k, i) for i, k in enumerate(seq_ref))
-#     inter = set(ind_dict_ref).intersection(seq_common)
-#     print(len(inter))
-#
-#     ind_dict_ref = [ind_dict_ref[x] for x in inter]
-#
-#     indices_common = dict((k, i) for i, k in enumerate(seq_common))
-#     indices_common = [indices_common[x] for x in inter]
-#
-#
-#     rt_ref = reference_unique[column_ref][ind_dict_ref].reset_index()
-#     rt_data = dataset_unique[column_dataset][indices_common].reset_index()
-#
-#     plt.scatter(rt_data[column_dataset].tolist(),rt_ref[column_ref].tolist(),s=0.1)
-#     plt.savefig('test.png')
-#
-#     xout, yout, wout = loess_1d(np.array(rt_data[column_dataset].tolist()), np.array(rt_ref[column_ref].tolist()),
-#                                 xnew=dataset[column_dataset],
-#                                 degree=1, frac=0.25,
-#                                 npoints=None, rotate=False, sigy=None)
-#
-#     plt.scatter(xout, yout, s=0.1)
-#     plt.savefig('test_2.png')
-#
-#     dataset[column_dataset] = yout
-#     return dataset
+def align(dataset, reference, column_dataset, column_ref, seq_data, seq_ref):
+    dataset_ref=dataset[dataset['state']=='train']
+    dataset_unique = dataset_ref[[seq_data,column_dataset]].groupby(seq_data).mean()
+    print('unique',len(dataset_unique))
+    reference_unique = reference[[seq_ref,column_ref]].groupby(seq_ref).mean()
+    seq_ref = reference_unique.index
+    seq_common = dataset_unique.index
+    seq_ref = seq_ref.tolist()
+    seq_common = seq_common.tolist()
+
+    seq_ref = [tuple(l) for l in seq_ref]
+    seq_common = [tuple(l) for l in seq_common]
+
+    ind_dict_ref = dict((k, i) for i, k in enumerate(seq_ref))
+    inter = set(ind_dict_ref).intersection(seq_common)
+    print(len(inter))
+
+    ind_dict_ref = [ind_dict_ref[x] for x in inter]
+
+    indices_common = dict((k, i) for i, k in enumerate(seq_common))
+    indices_common = [indices_common[x] for x in inter]
+
+
+    rt_ref = reference_unique[column_ref][ind_dict_ref].reset_index()
+    rt_data = dataset_unique[column_dataset][indices_common].reset_index()
+
+    plt.scatter(rt_data[column_dataset].tolist(),rt_ref[column_ref].tolist(),s=0.1)
+    plt.savefig('test.png')
+
+    xout, yout, wout = loess_1d(np.array(rt_data[column_dataset].tolist()), np.array(rt_ref[column_ref].tolist()),
+                                xnew=dataset[column_dataset],
+                                degree=1, frac=0.25,
+                                npoints=None, rotate=False, sigy=None)
+
+    plt.scatter(xout, yout, s=0.1)
+    plt.savefig('test_2.png')
+
+    dataset[column_dataset] = yout
+    return dataset
 
 def get_number_unique_peptide(dataset):
     seq = dataset['sequence']

diann_lib_processing.py → data/diann_lib_processing.py

@@ -2,14 +2,53 @@ import numpy as np
 import pandas as pb
 import pandas as pd
 import pyarrow.parquet as pq
+import pyarrow as pa
 import torch
+import matplotlib.pyplot as plt
 from model.model import ModelTransformer
 from config import load_args
 from data.dataset import load_data
 
+ALPHABET_UNMOD = {
+    "": 0,
+    "A": 1,
+    "C": 2,
+    "D": 3,
+    "E": 4,
+    "F": 5,
+    "G": 6,
+    "H": 7,
+    "I": 8,
+    "K": 9,
+    "L": 10,
+    "M": 11,
+    "N": 12,
+    "P": 13,
+    "Q": 14,
+    "R": 15,
+    "S": 16,
+    "T": 17,
+    "V": 18,
+    "W": 19,
+    "Y": 20,
+    "M(UniMod:35)": 21,
+    "CaC": 22
+}
+
+ALPHABET_UNMOD_REV = {v: k for k, v in ALPHABET_UNMOD.items()}
+
+def numerical_to_alphabetical_str(s):
+    seq = ''
+    s = s.replace('[','')
+    s = s.replace(']', '')
+    arr = s.split(',')
+    arr = list(map(int, arr))
+    for i in range(len(arr)):
+        seq+=ALPHABET_UNMOD_REV[arr[i]]
+    return seq
 
 def load_lib(path):
-    table = pq.read_table('data/spectral_lib/first_lib.parquet')
+    table = pq.read_table(path)
     table = table.to_pandas()
 
     return table
@@ -54,26 +93,54 @@ def predict(data_pred, model, output_path):
 
 
 if __name__ =='__main__':
-    # df = load_lib('data/spectral_lib/first_lib.parquet')
-    # df_2 = extract_sequence(df).reset_index(drop=True)
-    # df_2.to_csv('data/spectral_lib/data_uniprot_base.csv', index=False)
-
-
+    df = load_lib('spectral_lib/first_lib.parquet')
 
-    args = load_args()
+    plt.hist(df['RT'])
+    plt.savefig('test.png')
 
-    model = ModelTransformer(encoder_ff=args.encoder_ff, decoder_rt_ff=args.decoder_rt_ff,
-                             n_head=args.n_head, encoder_num_layer=args.encoder_num_layer,
-                             decoder_rt_num_layer=args.decoder_rt_num_layer, drop_rate=args.drop_rate,
-                             embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first, seq_length=30)
+    df_2 = pd.read_csv('data_prosit/data.csv')
 
-    if torch.cuda.is_available():
-        model = model.cuda()
+    plt.clf()
+    plt.hist(df_2['irt'])
+    plt.savefig('test2.png')
 
-    model.load_state_dict(torch.load(args.model_weigh, weights_only=True))
-
-    print(args.dataset_test)
-    data_test = load_data(data_source='data/spectral_lib/data_uniprot_base.csv', batch_size=args.batch_size, length=30, mode=args.split_test,
-                          seq_col=args.seq_test)
-
-    predict(data_test, model, args.output)
+    # df_2 = extract_sequence(df).reset_index(drop=True)
+    #
+    # pred = pd.read_csv('../output/out_uniprot_base.csv')
+    #
+    # pred['seq']=pred['seq'].map(numerical_to_alphabetical_str)
+    #
+    # pred['Modified.Sequence']=pred['seq']
\ No newline at end of file
+    #
+    # result = pd.merge(df,pred[['Modified.Sequence','rt pred']],on='Modified.Sequence',how='left')
+    #
+    # result['RT']=result['rt pred']
+    #
+    # result = result.drop('rt pred', axis=1)
+    #
+    # table = pa.Table.from_pandas(result)
+    #
+    # pq.write_table(table, 'spectral_lib/custom_first_lib.parquet')
+
+
+
+    # args = load_args()
+    #
+    # model = ModelTransformer(encoder_ff=args.encoder_ff, decoder_rt_ff=args.decoder_rt_ff,
+    #                          n_head=args.n_head, encoder_num_layer=args.encoder_num_layer,
+    #                          decoder_rt_num_layer=args.decoder_rt_num_layer, drop_rate=args.drop_rate,
+    #                          embedding_dim=args.embedding_dim, acti=args.activation, norm=args.norm_first, seq_length=30)
+    #
+    # if torch.cuda.is_available():
+    #     model = model.cuda()
+    #
+    # model.load_state_dict(torch.load(args.model_weigh, weights_only=True))
+    #
+    # print(args.dataset_test)
+    # data_test = load_data(data_source='data/spectral_lib/data_uniprot_base.csv', batch_size=args.batch_size, length=30, mode=args.split_test,
+    #                       seq_col=args.seq_test)
+    #
+    # predict(data_test, model, args.output)
+
+    plt.hist(df['RT'])
+    plt.savefig('test.png')
\ No newline at end of file

data/msp_file_extraction.py

+import pandas as pd
+
+
+if __name__ == '__main__':
+    seq=[]
+    file = open("spectral_lib/predicted_library.msp", "r")
+    content=file.readlines()
+    file.close()
+    for l in content :
+        if 'Name:'in l:
+            seq.append(l.split(':')[1].split('/')[0])
+
+    df = pd.DataFrame(seq,columns=['sequence'])
+    df['irt_scaled']=0
+    df['state'] = 'holdout'
+    df.to_csv('spectral_lib/df_predicted_library_oktoberfest.csv',index=False)
\ No newline at end of file