raise Exception("Model has never trained this n-gram: " + ngram) Exception: Model has never trained this n-gram: WNA

[devhimd19]

[kyu999]

Thank you for your report! The error means n-gram "WNA" is not trained because the corpus(uniprot trained one) does not contain such sequence, so you have to make your own corpus and train with it by yourself.

[devhimd19]

The corpus has the WNA. Can you please see the attached code and the input file.
output1.txt window_13re.txt biovec5.txt

I am getting the output but it is still showing the error

[AliASafdari]

Hi, @kyu999

I am facing the exact same error on my end too, but for the n-gram "KQE" instead.

Here's my code snippet -

pv = ProtVec('INPUT.FASTA', corpus_fname='OUTPUT.TXT', n=3) pv["QAT"] sequences = list(df[c]) (df[c] contains the AA sequence from which INPUT.FASTA was constructed) embeddings = [] for i in sequences: embed = pv.to_vecs(i) <- Error occurs here embeddings.append(embed)

Full code block, if it helps -

for d in data: df = pd.read_csv(d) dN = d[:-4] for c in cols: count = 1 with open('sequences_{a}_{b}.fasta'.format(a = c, b = dN), 'w') as f: for i in range(len(df)): print('>' + str(count) + '\n', df[c][i], file = f) count = count + 1 pv = ProtVec('sequences_{a}_{b}.fasta'.format(a = c, b = dN), corpus_fname='output_{a}_{b}.txt'.format(a = c, b = dN), n=3) pv["QAT"] sequences = list(df[c]) embeddings = [] for i in sequences: embed = pv.to_vecs(i) embeddings.append(embed) embedding = np.asarray(embeddings) all_embeddings = np.reshape(embedding, newshape=(embedding.shape[0], 300)) dF = pd.DataFrame(all_embeddings, columns = colN, dtype = object) dF['modification'] = df['modifications'] dF.to_csv('dataset-{a}_{b}.model'.format(a = c, b = dN)) pv.save('sequences_{a}_{b}.model'.format(a = c, b = dN))

(Idk why, but I can't seem to get this code block to indent properly.)

Please help me get past this error.