0.6.6 generate_pairs optimized for predict + minor doc edit

docs/source/usage.rst

@@ -9,13 +9,13 @@ Quick start
 SENSE-PPI can be used to predict pairwise physical interactions between proteins. The simplest input is single a FASTA file with protein sequences.
 The output is a .tsv file with all predictions as well as a secondary .tsv file that contains only positive interactions. By default, the predictions are made in "all vs all" manner: all possible protein pairs from the input file are considered.
 
-In order to copmute the predictions for all possible pairs from FASTA file, the following command can be used:
+In order to compute the predictions for all possible pairs from FASTA file, the following command can be used:
 
 .. code-block:: bash
 
     $ senseppi predict proteins.fasta
 
-By default, if no model is provided, the pre-trained model on human PPIs is used.
+By default, if no model is provided, the pre-trained model on human+worm+chicken+fly PPIs is used.
 
 List of commands
 ------------

senseppi/__init__.py

-__version__ = "0.6.5"
+__version__ = "0.6.6"
 __author__ = "Konstantin Volzhenin"
 
 from . import model, commands, esm2_model, dataset, utils, network_utils

senseppi/commands/predict.py

 from torch.utils.data import DataLoader
 import pytorch_lightning as pl
-from itertools import permutations, product
+from itertools import combinations
 import numpy as np
 import pandas as pd
 import pathlib
@@ -52,17 +52,16 @@ def generate_pairs(fasta_file, output_path, with_self=False):
     for record in SeqIO.parse(fasta_file, "fasta"):
         ids.append(record.id)
 
+    all_pairs = combinations(ids, 2)
+
     if with_self:
-        all_pairs = [p for p in product(ids, repeat=2)]
-    else:
-        all_pairs = [p for p in permutations(ids, 2)]
+        all_pairs = list(all_pairs)
+        for id in ids:
+            all_pairs.append((id, id))
 
-    pairs = []
-    for p in all_pairs:
-        if (p[1], p[0]) not in pairs and (p[0], p[1]) not in pairs:
-            pairs.append(p)
+    unique_pairs = set(all_pairs)
 
-    pairs = pd.DataFrame(pairs, columns=['seq1', 'seq2'])
+    pairs = pd.DataFrame(list(unique_pairs), columns=['seq1', 'seq2'])
 
     pairs.to_csv(output_path, sep='\t', index=False, header=False)