ENH update 09_Density

Author: cocodyq

General_Scripts/09_Density/01_100AA_copy_number_per_tax.py

@@ -1,21 +1,23 @@
 def store_100(infile):
-    import lzma
+    import gzip
     name = set()
-    with lzma.open(infile,'rt') as f:
+    with gzip.open(infile,'rt') as f:
         for line in f:
-            old,new = line.strip().split('\t')
-            name.add(old)
+            member,cluster = line.strip().split('\t')
+            name.add(member)
     return name
 
 def cal(infile,name,outfile):
     import lzma
+    
     kingdom = {}
     phylum = {}
     cl = {}
     order = {}
     family = {}
     genus = {}
     species = {}
+
     with lzma.open(infile,'rt') as f:
         for line in f:
             linelist = line.strip().split('\t',2)
@@ -75,7 +77,7 @@ def cal(infile,name,outfile):
         for key,value in species.items():
             out.write(f'{key}\t{value}\n')
 
-infile1 = '100AA_rename.tsv.xz'
+infile1 = 'GMSC.cluster.tsv.gz'
 infile2 = 'metag_cluster_taxonomy.tsv.xz'
 outfile = 'cpnumber_per_tax.tsv'
 

General_Scripts/09_Density/02_nbps_per_tax.py

@@ -1,12 +1,13 @@
 '''
+Concept:
 Add full ranks for npbs per taxon.
 '''
 def store_taxid_name(infile1):
     taxid_dict = {}
     with open(infile1,'rt') as f:
         for line in f:
             linelist = line.strip().split('\t',1)
-            if len(linelist) ==2 :
+            if len(linelist) == 2:
                 name = linelist[1].replace('d__','').replace('p__','').replace('c__','').replace('o__','').replace('f__','').replace('g__','').replace('s__','').replace('\t',';')
                 taxid_dict[linelist[0]] = name
     return taxid_dict
@@ -43,6 +44,7 @@ def per_tax(infile,outfile):
     family_number = {}
     genus_number = {}
     species_number = {}
+    
     with open(infile,'rt') as f:
         for line in f:
             taxid,rank,name,nbps = line.strip().split('\t')

General_Scripts/09_Density/03_calculate_density.py

@@ -1,4 +1,5 @@
 '''
+Concept:
 Calculate density of phylum and genus
 '''
 import pandas as pd

General_Scripts/09_Density/Readme.md

@@ -2,6 +2,6 @@
 
 | **Code** | **Description** | **Input** | **Output** |
 | :---: | :---: | :---: | :---: |
-| 01_100AA_copy_number_per_tax.py | Calculate the copy number of smORFs per taxonomy | 100AA_rename.tsv.xz metag_cluster_taxonomy.tsv.xz | cpnumber_per_tax.tsv |
+| 01_100AA_copy_number_per_tax.py | Calculate the copy number of smORFs per taxonomy | GMSC.cluster.tsv.gz metag_cluster_taxonomy.tsv.xz | cpnumber_per_tax.tsv |
 | 02_nbps_per_tax.py | Calculate nbps per taxonomy | taxid_fullname_gtdb.tsv bps-per-taxon.tsv | per_tax_rank.txt full_nbp.txt |
 | 03_calculate_density.py | Calculate density of phylum and genus | cpnumber_per_tax.tsv per_tax_rank.txt| density_phylum.tsv density_genus.tsv |