In [2]:

    

!ls ../../1_find_poophage/2_hmmer/


    

    




1_extract_poophage_reads.sh	454_seqs_poophage_PfamA.tbl
2_hmmscan_to_pfam.sh		454_seqs_poophage_pVOG.domtbl
2_hmmscan_to_pvog.sh		454_seqs_poophage_pVOG.out
2_hmmscan_to_vogdb.sh		454_seqs_poophage_pVOG.tbl
454_seqs_poophage.fa		454_seqs_poophage_vogdb.domtbl
454_seqs_poophage_PfamA.domtbl	454_seqs_poophage_vogdb.out
454_seqs_poophage_PfamA.out	454_seqs_poophage_vogdb.tbl

In [9]:

    

import pandas as pd


    

In [26]:

    

#1. Load poophage seqids
poophage_seqs = []
with open("../../1_find_poophage/2_hmmer/454_seqs_poophage.fa") as fh:
    for line in fh:
        if line.startswith(">"):
            poophage_seqs.append(line.rstrip("\n").lstrip(">"))
            
print(len(poophage_seqs))
poophage_seqs = frozenset(poophage_seqs)


    

    




1147

In [23]:

    

kaiju_virus_df = pd.read_csv("./454_seqs_kaiju.names.virus.txt",sep="\t",header=None,names=["classified","seqid","taxid","tax"])
kaiju_virus_df.head()


    

    
Out[23]:






  

    

      

      
classified
      
seqid
      
taxid
      
tax
    
  
  

    

      
0
      
C
      
FTSPZO101CHNQ7
      
196894
      
Viruses; Siphoviridae; NA;
    
    

      
1
      
C
      
DBA-SLE_c5570
      
1341019
      
Viruses; Parvoviridae; Parvovirus NIH-CQV;
    
    

      
2
      
C
      
GB3LKKR01ED1RA
      
1792245
      
Viruses; Myoviridae; Bacillus virus Deepblue;
    
    

      
3
      
C
      
GB3LKKR01DUZ77
      
948870
      
Viruses; Myoviridae; Enterobacteria phage phi92;
    
    

      
4
      
C
      
FTSPZO101E24BN
      
1608451
      
Viruses; NA; Phytophthora parasitica virus;
    
  

In [31]:

    

#It's empty !
kaiju_virus_df[kaiju_virus_df["seqid"].apply(lambda x: x in poophage_seqs)].shape


    

    
Out[31]:





(0, 4)

In [ ]: