Added protparam script to calculate protein length, Mw, pI and aa content

Author: Evgenii Osipov

scripts/protparam.py

@@ -0,0 +1,61 @@
+from pymol import cmd
+from io import StringIO
+
+try:
+    from Bio.SeqUtils.ProtParam import ProteinAnalysis
+    from Bio import SeqIO
+    from Bio.Seq import Seq
+except ModuleNotFoundError:
+    # Note that Bio package might be missing from Pymol 2 installation!
+    # run "pip install biopython"
+    import subprocess, sys
+    print("Biopython is missing! Installing...")
+    try:
+        subprocess.check_call([sys.executable, "-m", "pip", "install", 'biopython'])
+        print(f"Successfully installed biopython")
+        # need to import libs after first installation, otherwise protparam fails to run
+        from Bio.SeqUtils.ProtParam import ProteinAnalysis
+        from Bio import SeqIO
+        from Bio.Seq import Seq
+    except subprocess.CalledProcessError as e:
+        print(f"Failed to install biopython: {e}")
+
+@cmd.extend
+def protparam(selection='enabled', bychain=0):
+    '''
+    DESCRIPTION:
+    Given selection, calculates common protein properties, like Mw, pI, length and aminoacid content.
+    By default, combines all chains of each object into the single sequence.
+    
+    USAGE:
+    protparam selection, [bychain]
+
+    DEPENDENCIES:
+    biopython
+    '''
+    #TODO: add pretty output suitable for copy-pasting
+    for entry in cmd.get_object_list(selection):
+        sequence_obj = cmd.get_fastastr(f"{selection} and {entry}")
+        fasta_io = StringIO(sequence_obj)
+        sequences = list(SeqIO.parse(fasta_io, "fasta"))
+        if not bychain:
+            #combine all chains into the single one
+            sequences = [Seq('').join([s.seq for s in sequences])]
+        else:
+            sequences = [s.seq for s in sequences]
+        for sequence in sequences:
+            sequence = str(sequence).replace('?','').strip()
+            analysis = ProteinAnalysis(sequence)
+            counts_aa = analysis.count_amino_acids()
+            print(f"Protein name: {entry}")
+            print(f"Sequence: {sequence}")
+            print(f"\nProtein length: {analysis.length} aa")
+            print(f"Molecular Weight: {analysis.molecular_weight():.1f} Da")
+            print(f"Isoelectric point: {analysis.isoelectric_point():.2f}")
+            print(f"Count of aminoacids: {counts_aa}\n\n")
+
+
+def test_protparam():
+    cmd.reinitialize()
+    cmd.fab('ACDEFG?HIKLMN', 'm1')
+    cmd.do('protparam m1')