Merge branch 'stage_aglae' of ssh://forge.ibisc.univ-evry.fr:6622/lbecquey/RNANet into stage_aglae

Former-commit-id: 2ee1c748

@@ -201,7 +201,8 @@ class Chain:
             self.file = path_to_3D_data+"rna_mapped_to_Rfam/"+self.chain_label+".cif"
             self.file = path_to_3D_data+"rna_mapped_to_Rfam/"+self.chain_label+".cif"
         else:
         else:
             status = f"Extract {self.pdb_id}-{self.pdb_chain_id}"
             status = f"Extract {self.pdb_id}-{self.pdb_chain_id}"
-            self.file = path_to_3D_data+"rna_only/"+self.chain_label+".cif"
+            self.file = path_to_3D_data+"renumbered_rna_only/"+self.chain_label+".cif"
+            #self.file = path_to_3D_data+"rna_only/"+self.chain_label+".cif"
 
 
         # Check if file exists, if yes, abort (do not recompute)
         # Check if file exists, if yes, abort (do not recompute)
         if os.path.exists(self.file):
         if os.path.exists(self.file):
@@ -235,11 +236,85 @@ class Chain:
 
 
             # Define a selection
             # Define a selection
             sel = SelectivePortionSelector(model_idx, self.pdb_chain_id, valid_set, khetatm)
             sel = SelectivePortionSelector(model_idx, self.pdb_chain_id, valid_set, khetatm)
-
+            
+            # save the selection sel into a new structure
+            new_s=pdb.Structure.Structure(s.get_id())
+    
+            for model in s:
+                if sel.accept_model(model):
+                    new_model=pdb.Model.Model(model.get_id())
+                    for chain in model:
+                        if sel.accept_chain(chain):
+                            new_chain=pdb.Chain.Chain(chain.get_id())
+                            for res in chain:
+                                if sel.accept_residue(res):
+                                    res_atoms=res.get_atoms()
+                                    new_residu=pdb.Residue.Residue(res.get_id(), res.get_resname(), res.get_segid())
+                                    for atom in list(res.get_atoms()):
+                                        if sel.accept_atom(atom):
+                                            new_atom=atom.copy()
+                                            new_residu.add(new_atom)
+                                    new_chain.add(new_residu)
+
+                            new_model.add(new_chain)
+                    new_s.add(new_model)
+
+            # renumber this structure (portion of the original) with the index_chain and save it in a cif file
+            t=pdb.Structure.Structure(new_s.get_id())
+            model=new_s[0]
+            new_model_t=pdb.Model.Model(model.get_id())
+            for chain in model:
+                nums=df[["index_chain", "old_nt_resnum", "nt_name"]]
+                new_chain_t=pdb.Chain.Chain(chain.get_id())
+                for i in nums.index:
+                    resseq=nums.at[i, 'old_nt_resnum']
+                    icode_res=' '
+                    if type(resseq) is str:
+                        if resseq=='not resolved':
+                            continue
+                        if resseq[0] != '-' :
+                            while resseq.isdigit() is False:
+                                l=len(resseq)
+                                if icode_res==' ':
+                                    icode_res=resseq[l-1]
+                                else :
+                                    icode_res=resseq[l-1]+icode_res
+                                resseq=resseq[:l-1]
+                    resseq=int(resseq)
+                    index_chain=nums.at[i, "index_chain"]
+                    nt=nums.at[i, "nt_name"]
+                    if nt == 'A' or nt == 'G' or nt == 'C' or nt == 'U':
+                        res=chain[(' ', resseq, icode_res)]
+                    else : #modified nucleotides (e.g. chain 5l4o_1_A)
+                        het='H_' + nt
+                        res=chain[(het, resseq, icode_res)]
+                    res_id=res.get_id()
+                    res_id=list(res_id)
+                    res_id[1]=index_chain
+                    res_id[2]=' '
+                    res_id[0]=' '
+                    res_id=tuple(res_id)
+                    res_atoms=res.get_atoms()
+                    new_residu_t=pdb.Residue.Residue(res_id, res.get_resname(), res.get_segid())
+                    for atom in list(res.get_atoms()):
+                        new_atom_t=atom.copy()
+                        new_residu_t.add(new_atom_t)
+                    new_chain_t.add(new_residu_t)
+
+                new_model_t.add(new_chain_t)
+            t.add(new_model_t)
+                    
+            # Save that renumbered selection on the mmCIF object s to file       
+            ioobj = pdb.MMCIFIO()
+            ioobj.set_structure(t)
+            ioobj.save(self.file)   
+            
             # Save that selection on the mmCIF object s to file
             # Save that selection on the mmCIF object s to file
+            '''
             ioobj = pdb.MMCIFIO()
             ioobj = pdb.MMCIFIO()
             ioobj.set_structure(s)
             ioobj.set_structure(s)
             ioobj.save(self.file, sel)
             ioobj.save(self.file, sel)
+            '''
 
 
         notify(status)
         notify(status)
 
 
@@ -1311,9 +1386,14 @@ class Pipeline:
             if self.HOMOLOGY and not os.path.isdir(path_to_3D_data + "rna_mapped_to_Rfam"):
             if self.HOMOLOGY and not os.path.isdir(path_to_3D_data + "rna_mapped_to_Rfam"):
                 # for the portions mapped to Rfam
                 # for the portions mapped to Rfam
                 os.makedirs(path_to_3D_data + "rna_mapped_to_Rfam")
                 os.makedirs(path_to_3D_data + "rna_mapped_to_Rfam")
+            '''
             if (not self.HOMOLOGY) and not os.path.isdir(path_to_3D_data + "rna_only"):
             if (not self.HOMOLOGY) and not os.path.isdir(path_to_3D_data + "rna_only"):
                 # extract chains of pure RNA
                 # extract chains of pure RNA
                 os.makedirs(path_to_3D_data + "rna_only")
                 os.makedirs(path_to_3D_data + "rna_only")
+            '''
+            if (not self.HOMOLOGY) and not os.path.isdir(path_to_3D_data + "renumbered_rna_only"):
+                # extract chains of pure RNA
+                os.makedirs(path_to_3D_data + "renumbered_rna_only")
 
 
         # define and run jobs
         # define and run jobs
         joblist = []
         joblist = []
@@ -2809,8 +2889,7 @@ def work_save(c, homology=True):
             
             
         filename = path_to_3D_data + "datapoints/" + c.chain_label
         filename = path_to_3D_data + "datapoints/" + c.chain_label
     conn.close()
     conn.close()
-    pairs=df[['index_chain', 'old_nt_resnum', 'paired']]
-    print(pairs.iloc[:40])
+
     df.to_csv(filename, float_format="%.2f", index=False)
     df.to_csv(filename, float_format="%.2f", index=False)
 
 
 if __name__ == "__main__":
 if __name__ == "__main__":

+f665319aeeca9246a8617b209cdf02646f547518
\ No newline at end of file

@@ -1710,3 +1710,480 @@
 6pmi_1_3
 6pmi_1_3
 6pmj_1_3
 6pmj_1_3
 5hjz_1_C
 5hjz_1_C
+6vff_1_D+6VFF
+5lr4_1_A+5LR4
+3j0o_1_E+3J0O
+6z8k_1_P+6Z8K
+5kvj_1_A+5KVJ
+3lrr_1_C+3LRR
+4b3r_1_Z+4B3R
+1yzd_1_A+1YZD
+7dco_1_H+7DCO
+2gjw_1_I+2GJW
+7bv2_1_T+7BV2
+6nd4_1_0+6ND4
+3r1e_1_A+3R1E
+5hnj_1_A+5HNJ
+6sgc_1_54+6SGC
+2uxc_1_X+2UXC
+2g3s_1_I+2G3S
+6bu9_1_B+6BU9
+205d_1_A+205D
+1n32_1_Y+1N32
+6s8b_1_V+6S8B
+4wta_1_T+4WTA
+1i9x_1_A+1I9X
+5xy3_1_1+5XY3
+7nun_1_C
+7l1f_1_T+7L1F
+3p59_1_E+3P59
+5e3h_1_B+5E3H
+5m3h_1_V+5M3H
+5tbw_1_1+5TBW
+3nl0_1_B+3NL0
+6qdv_1_2+6QDV
+7ctt_1_T+7CTT
+5f8j_1_B+5F8J
+3bbm_1_B+3BBM
+4wtm_1_T+4WTM
+6hct_1_A+6HCT
+1kd4_1_A+1KD4
+6swa_1_Q+6SWA
+1qc0_1_C+1QC0
+6zmi_1_L5+6ZMI
+2p7e_1_D+2P7E
+2uxb_1_Y+2UXB
+4v8p_1_D1+4V8P
+6wb0_1_D+6WB0
+1yyw_1_I+1YYW
+4jab_1_A+4JAB
+439d_1_A+439D
+3bnr_1_C+3BNR
+6yyt_1_P+6YYT
+5c9h_1_C+5C9H
+7jqq_1_O
+3fte_1_C+3FTE
+2zi0_1_C+2ZI0
+7nuq_1_C
+4k31_1_B+4K31
+6mpi_1_X+6MPI
+7jl1_1_X+7JL1
+3dh3_1_F+3DH3
+6ifk_1_N+6IFK
+4wtd_1_T+4WTD
+7cyq_1_J+7CYQ
+1rpu_1_C+1RPU
+3c3z_1_A+3C3Z
+6lqp_1_SA+6LQP
+3jb7_1_T+3JB7
+5jbj_1_X+5JBJ
+3j0o_1_G+3J0O
+3ciy_1_C+3CIY
+7ndj_1_C
+419d_1_C+419D
+6zj3_1_LH+6ZJ3
+4p43_1_B+4P43
+1yyk_1_C+1YYK
+1iha_1_A+1IHA
+4v7e_1_AA+4V7E
+7jl3_1_X+7JL3
+3szx_1_A+3SZX
+4wtl_1_T+4WTL
+1xnq_1_X+1XNQ
+5d8t_1_A+5D8T
+4wrt_1_V+4WRT
+1z7f_1_B+1Z7F
+4e58_1_D+4E58
+3l25_1_C+3L25
+3vnv_1_T+3VNV
+1fuf_1_A+1FUF
+5ed2_1_F+5ED2
+5y6z_1_B+5Y6Z
+5nrl_1_2+5NRL
+2gxb_1_E+2GXB
+7m4y_1_A
+4wtj_1_T+4WTJ
+5jxs_1_B+5JXS
+2nug_1_E+2NUG
+6lsf_1_C+6LSF
+3cgp_1_B+3CGP
+4qpx_1_P+4QPX
+3avy_1_T+3AVY
+6hak_1_T+6HAK
+6z6b_1_HHH+6Z6B
+6o5f_1_D+6O5F
+3siv_1_C+3SIV
+4z4d_1_B+4Z4D
+3td0_1_B+3TD0
+5lr5_1_C+5LR5
+4kyy_1_A+4KYY
+7am2_1_R1
+7d8c_1_C
+7d8c_1_B
+359d_1_B+359D
+6mkn_1_X+6MKN
+246d_1_A+246D
+4wte_1_T+4WTE
+4u35_1_B+4U35
+2v6w_1_A+2V6W
+4eya_1_E+4EYA
+2e9r_1_A+2E9R
+1jzv_1_B+1JZV
+3r1c_1_Q+3R1C
+5jch_1_U+5JCH
+1t0e_1_A+1T0E
+4a36_1_R+4A36
+3r1d_1_A+3R1D
+6jq5_1_B+6JQ5
+1t0k_1_D+1T0K
+6qct_1_R+6QCT
+6z1p_1_AB+6Z1P
+7krn_1_T+7KRN
+1j9h_1_A+1J9H
+3ftf_1_C+3FTF
+6b19_1_C+6B19
+6v4x_1_Z+6V4X
+438d_1_A+438D
+6rxu_1_C1+6RXU
+1xnr_1_X+1XNR
+5ed1_1_C+5ED1
+3mei_1_A+3MEI
+3htx_1_B+3HTX
+4eya_1_G+4EYA
+4e48_1_A+4E48
+2r1s_1_A+2R1S
+6cap_1_W+6CAP
+5lr3_1_A+5LR3
+5gan_1_V+5GAN
+5dhc_1_C+5DHC
+3cgr_1_B+3CGR
+353d_1_A+353D
+6lqs_1_3A+6LQS
+6htu_1_D+6HTU
+3jb9_1_N+3JB9
+6tz2_1_N+6TZ2
+6rxx_1_C1+6RXX
+4ts2_1_Y+4TS2
+4e59_1_A+4E59
+1csl_1_B+1CSL
+6waz_1_D+6WAZ
+4xw0_1_A+4XW0
+2oe5_1_B+2OE5
+7dvq_1_F+7DVQ
+6gkh_1_X+6GKH
+2noq_1_D+2NOQ
+7k5l_1_R
+6ra4_1_L+6RA4
+6v9b_1_B+6V9B
+6ozo_1_C+6OZO
+4gl2_1_C+4GL2
+2i91_1_F+2I91
+6mdz_1_E+6MDZ
+5ay2_1_A+5AY2
+1zci_1_A+1ZCI
+1nuv_1_E+1NUV
+5wnt_1_B+5WNT
+6szv_1_R+6SZV
+364d_1_C+364D
+2oe8_1_B+2OE8
+4kq0_1_B+4KQ0
+3r1c_1_K+3R1C
+2h1m_1_A+2H1M
+6z1p_1_BB+6Z1P
+5zwo_1_H+5ZWO
+6az3_1_1+6AZ3
+2et8_1_A+2ET8
+1feu_1_C+1FEU
+2uxd_1_Y+2UXD
+5g4t_1_A+5G4T
+3ol9_1_B+3OL9
+3adi_1_D+3ADI
+7b3d_1_T+7B3D
+5dv7_1_A+5DV7
+5w0o_1_C+5W0O
+2g92_1_A+2G92
+6yhs_1_8+6YHS
+4j50_1_A+4J50
+3pla_1_G+3PLA
+1ry1_1_Q+1RY1
+2o3x_1_A+2O3X
+4x62_1_B+4X62
+1nta_1_A+1NTA
+1zx7_1_C+1ZX7
+6id1_1_H+6ID1
+2f8s_1_C+2F8S
+472d_1_A+472D
+5amr_1_B+5AMR
+2pn3_1_A+2PN3
+2e9t_1_B+2E9T
+3fs0_1_B+3FS0
+3j9b_1_K+3J9B
+1nuv_1_A+1NUV
+1yz9_1_C+1YZ9
+4ig8_1_B+4IG8
+3s49_1_E+3S49
+7krp_1_P+7KRP
+1h1k_1_K+1H1K
+6ia2_1_A+6IA2
+4wtk_1_T+4WTK
+4gha_1_I+4GHA
+7jju_1_A+7JJU
+1duq_1_B+1DUQ
+4db2_1_E+4DB2
+1mhk_1_L+1MHK
+6ozr_1_C+6OZR
+4o41_1_C+4O41
+6k32_1_T+6K32
+6vem_1_A+6VEM
+6gvy_1_C+6GVY
+3loa_1_B+3LOA
+4k0k_1_Y+4K0K
+1h1k_1_I+1H1K
+4wtc_1_T+4WTC
+1ry1_1_O+1RY1
+4kz2_1_B+4KZ2
+4wzq_1_B+4WZQ
+3tzr_1_A+3TZR
+3nvk_1_K+3NVK
+1g1x_1_J+1G1X
+2v7r_1_A+2V7R
+1n35_1_C+1N35
+6m6s_1_C+6M6S
+3vyx_1_B+3VYX
+1tfw_1_G+1TFW
+1sdr_1_A+1SDR
+1yvp_1_C+1YVP
+6zj3_1_LC+6ZJ3
+5dhb_1_A+5DHB
+1ytu_1_E+1YTU
+5t5h_1_A+5T5H
+435d_1_C+435D
+3cgs_1_B+3CGS
+4phy_1_A+4PHY
+1dqh_1_B+1DQH
+5h0r_1_H+5H0R
+3kmq_1_B+3KMQ
+7m4y_1_V
+6hct_1_E+6HCT
+6ty9_1_T+6TY9
+5ued_1_A+5UED
+1qbp_1_A+1QBP
+6bgb_1_A+6BGB
+6kur_1_V+6KUR
+464d_1_A+464D
+5l00_1_A+5L00
+485d_1_A+485D
+4v6w_1_A5+4V6W
+422d_1_A+422D
+5c45_1_X+5C45
+1rna_1_A+1RNA
+5xxb_1_1+5XXB
+3r1c_1_E+3R1C
+4zc7_1_A+4ZC7
+5lqo_1_A+5LQO
+4v7h_1_B5+4V7H
+4gkk_1_X+4GKK
+6ogz_1_B+6OGZ
+3jxq_1_L+3JXQ
+7kjv_1_C+7KJV
+6rr7_1_D+6RR7
+6xqb_1_F+6XQB
+5jaj_1_X+5JAJ
+7bzf_1_F+7BZF
+5wnp_1_A+5WNP
+3avv_1_T+3AVV
+2zko_1_C+2ZKO
+3avw_1_T+3AVW
+1kd5_1_A+1KD5
+1zh5_1_C+1ZH5
+2val_1_A+2VAL
+1dqf_1_B+1DQF
+7b9v_1_2+7B9V
+1icg_1_A+1ICG
+4k4s_1_B+4K4S
+4fvu_1_B+4FVU
+3h5x_1_T+3H5X
+4ngf_1_G+4NGF
+3nmu_1_E+3NMU
+2a0p_1_A+2A0P
+1d4r_1_B+1D4R
+4rwn_1_B+4RWN
+3j7q_1_5+3J7Q
+4mcf_1_B+4MCF
+3avx_1_T+3AVX
+7bg6_1_C
+4gv6_1_C+4GV6
+3mqk_1_D+3MQK
+5zuu_1_E+5ZUU
+6d06_1_B+6D06
+5lqt_1_A+5LQT
+7am2_1_R2+7AM2
+6zqa_1_D3+6ZQA
+4gkj_1_X+4GKJ
+7c2k_1_F+7C2K
+7aap_1_T+7AAP
+1sa9_1_C+1SA9
+6sx2_1_C+6SX2
+5vci_1_A+5VCI
+280d_1_A+280D
+4yhw_1_F+4YHW
+6pzq_1_J+6PZQ
+3gm7_1_G+3GM7
+5c9h_1_G+5C9H
+6lse_1_C+6LSE
+3cz3_1_G+3CZ3
+7b3c_1_T+7B3C
+5une_1_A+5UNE
+2e9z_1_B+2E9Z
+4k4z_1_N+4K4Z
+6szu_1_V+6SZU
+5ew4_1_B+5EW4
+3ncu_1_C+3NCU
+2vuq_1_A+2VUQ
+6hu6_1_B+6HU6
+5v1k_1_A+5V1K
+3dkn_1_D+3DKN
+2xsl_1_C+2XSL
+6e9f_1_B+6E9F
+418d_1_A+418D
+4e6b_1_A+4E6B
+6g19_1_X+6G19
+1wne_1_B+1WNE
+4jrt_1_A+4JRT
+2ao5_1_C+2AO5
+6oj6_1_T+6OJ6
+2r92_1_T+2R92
+4k4u_1_B+4K4U
+4u3l_1_A+4U3L
+1tfw_1_H+1TFW
+6kuk_1_V+6KUK
+402d_1_A+402D
+3og8_1_C+3OG8
+4yhw_1_E+4YHW
+3cjz_1_A+3CJZ
+6bbo_1_F+6BBO
+6mfn_1_E+6MFN
+5u0q_1_C+5U0Q
+2ec0_1_B+2EC0
+3glp_1_B+3GLP
+5h1k_1_C+5H1K
+4dr6_1_W+4DR6
+4eya_1_M+4EYA
+3hjy_1_D+3HJY
+3kna_1_B+3KNA
+3vnu_1_G+3VNU
+6tw1_1_V+6TW1
+6lsg_1_B+6LSG
+3ks8_1_E+3KS8
+4e6b_1_E+4E6B
+406d_1_A+406D
+1qc0_1_B+1QC0
+3a6p_1_E+3A6P
+6kuv_1_V+6KUV
+6qwl_1_W+6QWL
+4wsb_1_V+4WSB
+332d_1_A+332D
+3qrp_1_B+3QRP
+1qcu_1_A+1QCU
+4p3u_1_C+4P3U
+4e78_1_P+4E78
+4eya_1_O+4EYA
+3gvn_1_A+3GVN
+3j0e_1_E+3J0E
+6qn3_1_A+6QN3
+2az0_1_C+2AZ0
+4wzm_1_B+4WZM
+2o3y_1_A+2O3Y
+4u34_1_B+4U34
+377d_1_A+377D
+4z31_1_C+4Z31
+5wnu_1_B+5WNU
+2bgg_1_P+2BGG
+3avu_1_T+3AVU
+6t0s_1_V+6T0S
+6dn2_1_Y+6DN2
+5f8n_1_B+5F8N
+7d3j_1_B
+6kuj_1_V+6KUJ
+6bk8_1_2+6BK8
+5g4u_1_I+5G4U
+2rfk_1_E+2RFK
+3bnn_1_C+3BNN
+6g1x_1_X+6G1X
+2o3v_1_A+2O3V
+1dfu_1_M+1DFU
+1tfy_1_H+1TFY
+1r3e_1_D+1R3E
+6tz1_1_T+6TZ1
+1uon_1_C+1UON
+4eya_1_K+4EYA
+4g9z_1_E+4G9Z
+2g91_1_A+2G91
+6o7h_1_G+6O7H
+3nma_1_B+3NMA
+3wbm_1_X+3WBM
+1tfy_1_G+1TFY
+4wti_1_T+4WTI
+3sj2_1_A+3SJ2
+4v8m_1_BA+4V8M
+6zqd_1_D4+6ZQD
+6o1o_1_M+6O1O
+3adl_1_B+3ADL
+2w89_1_A+2W89
+433d_1_A+433D
+3r2d_1_R+3R2D
+6ozp_1_C+6OZP
+6lsh_1_B+6LSH
+5t3k_1_A+5T3K
+6t0v_1_R+6T0V
+409d_1_C+409D
+4e7a_1_P+4E7A
+3ftm_1_D+3FTM
+3koa_1_B+3KOA
+7abg_1_2+7ABG
+1saq_1_A+1SAQ
+4p3u_1_B+4P3U
+6n4o_1_C+6N4O
+354d_1_A+354D
+4pco_1_A+4PCO
+4a2x_1_L+4A2X
+1n33_1_Y+1N33
+6dti_1_X+6DTI
+5cnr_1_A+5CNR
+1qcu_1_C+1QCU
+7cxm_1_J+7CXM
+1zz5_1_A+1ZZ5
+4x2b_1_B+4X2B
+4p3t_1_A+4P3T
+4jah_1_A+4JAH
+4e5c_1_A+4E5C
+1n38_1_C+1N38
+6qit_1_A+6QIT
+2f8t_1_C+2F8T
+1ry1_1_N+1RY1
+6mpf_1_X+6MPF
+397d_1_A+397D
+5tsn_1_P+5TSN
+5jcf_1_Y+5JCF
+6nta_1_XV+6NTA
+7dmq_1_B+7DMQ
+3eqt_1_C+3EQT
+4p97_1_A+4P97
+6ifl_1_J+6IFL
+315d_1_A+315D
+5zwm_1_H+5ZWM
+6exn_1_2+6EXN
+3nj7_1_C+3NJ7
+5jc7_1_X+5JC7
+5lqw_1_6+5LQW
+5yzg_1_H+5YZG
+3j79_1_A+3J79
+4qik_1_C+4QIK
+4erd_1_C+4ERD
+6uz7_1_5+6UZ7
+1h1k_1_J+1H1K
+2xbm_1_E+2XBM
+1jbr_1_C+1JBR
+2q1o_1_C+2Q1O
+1lnt_1_A+1LNT
+1g1x_1_D+1G1X