Papyrus README file =================== last modified: Aug. 25, 2022 Papyrus version: 5.5 Total number of files: 14 =================== Papyrus is an aggregated dataset of small molecule bioactivities. The very high quality Papyrus++ data in contained in the 05.5++_combined_set_without_stereochemistry.tsv.xz file. The other data is split into 3 tab-separated files: - 05.5_combined_set_without_stereochemistry.tsv.xz This file contains the highly standardized and normalized data. Although not containing stereochemistry, it provides the highest quality of the data. This file can be readily used. - 05.5_combined_set_with_stereochemistry.tsv.xz This file contains standardized and normalized activity values but molecules are not standardized. It is of lower confidence than the data above although containing stereochemistry. By "molecules are not standardized" we mean that sources have applied different standardization processes and the reported stereochemistry may vary for the molecule the activities relate to. It is recommended to standardize the data before using it. - 05.5_combined_set_protein_targets.tsv.xz This file contains data about protein targets (e.g. sequence, organism, classification). In addition to these, molecular structures are available in SDfile format: - 05.5_combined_2D_set_without_stereochemistry.sd.xz This file contains the unique molecules found in 05.5_combined_set_without_stereochemistry.tsv.xz - 05.5_combined_3D_set_with_stereochemistry.xd.xz This file contains the unique molecules found in 05.5_combined_set_with_stereochemistry.tsv.xz Description of the data in Papyrus: The 05.5_combined_set_protein_targets.tsv.xz file contains the following colunms: - target_id A unique Papyrus protein identifier. It results from the concatenation of accessions and mutations (e.g. P47747_WT or P10721_V559D_T670I). - HGNC_symbol The HGNC gene symbol if could be found. - UniProtID The UniProt identifier of the sequence. - Status Status of the UniProt entry, either 'reviewed' if entry comes from SwissProt or 'unreviewed' if comes from TrEMBL. - Organism Organism of the protein. - Classification Protein classification as given by ChEMBL (version 30). Levels are separated by "->". Multiple classifications are separated by a semilcolon (";") - Length Length of the protein sequence. - Sequence Protein sequence including mutations. The 05.5_combined_set_without_stereochemistry.tsv.xz file contains: - Activity_ID A unique Papyrus compound-protein identifier. It results from the concatenation of the molcule InChI and protein unique identifier. - Quality quality assigned to the activity data - source Sources of the activity values separeted by semilcolons. - CID Compound identifiers (only original identifiers for ExCAPE-DB and ChEMBL data). - SMILES Standardized SMILES notation of the molecule. - connectivity Connectivity of the molecule. - InChIKey InChIKey of the molecule. - InChI InChI of the molecule. - InChI_AuxInfo InChI_AuxInfo of the molecule. - target_id Unique protein identifier, corresponds to the "target_id" in 05.4_combined_set_protein_targets.tsv.xz. - accession Protein UniProt Accession - Protein_Type Either "WT" for wild-type sequences or a sequence of the mutations underscore-separated. - AID Assay identifiers (only original identifiers for ExCAPE-DB and ChEMBL data). - doc_id First published or filed document characterizing the compound-protein interation. This does not always correspond to the source of the reported activities. - Year Year the document referred to by "doc_id" was published or filed. - all_docs_id List of documents values in "pchembl_value" originate from. - all_years List of years the documents in "all_docs_id" were filed or published. - type_IC50 Binary indicator of the types of the values in "pchembl_value" semicolon-separated. If 0 then the activity is not an IC50, if 1 then it is. If a unique value, then all activities have the same type_IC50. - type_EC50 Binary indicator of the types of the values in "pchembl_value" semicolon-separated. If 0 then the activity is not an EC50, if 1 then it is. If a unique value, then all activities have the same type_EC50. - type_KD Binary indicator of the types of the values in "pchembl_value" semicolon-separated. If 0 then the activity is not an KD, if 1 then it is. If a unique value, then all activities have the same type_KD. - type_Ki Binary indicator of the types of the values in "pchembl_value" semicolon-separated. If 0 then the activity is not an Ki, if 1 then it is. If a unique value, then all activities have the same type_Ki. - type_other Binary indicator of the types of the values in "pchembl_value" semicolon-separated. If 0 then the activity is IC50, EC50, Ki or KD, if 1 then it is not. If a unique value, then all activities have the same type_other. - Activity_class Binary indication of activity if "pchembl_value" is empty. - relation Comparison sign to indicate if "pchembl_value" is exact ("=") or censored ("<", ">", "<=", ">="). - pchembl_value Ensemble of log-transformed activity values available for then compound-protein pair at the highest quality possible. - pchembl_value_Mean Mean average of "pchembl_value" activities. - pchembl_value_StdDev Standard deviation of "pchembl_value" activities. - pchembl_value_SEM Standard error of the mean of "pchembl_value" activities. - pchembl_value_N Number of "pchembl_value" activities for that compound-protein pair. - pchembl_value_Median Median of "pchembl_value" activities. - pchembl_value_MAD Mean absolute error of "pchembl_value" activities. The 05.5_combined_set_with_stereochemistry.tsv.xz file contains the same columns as the 05.5_combined_set_without_stereochemistry.tsv.xz file with few exceptions: - connectivity is not present - SMILES is not standardized but contains stereochemistry. The 05.5_combined_2D_set_without_stereochemistry.sd.xz file contains connection tables of molecules with 2D coordinates and few properties: - SMILES Standardized SMILES notation of the molecule in the related tab-separated file. - connectivity Connectivity of the molecule that should be used to join this file with its related tab-separated file. - InChIKey InChIKey of the molecule. - NEMAKey NEMAKey of the molecule (internally used to identify molecules with same connectivity but differing tautomeric state). The 05.5_combined_3D_set_with_stereochemistry.sd.xz file contains connection tables of molecules with 3D coordinates and the same properties as the 05.5_combined_2D_set_without_stereochemistry.sd.xz filewith few axception. - connectivity is no present. InChIKey should be used in order to join this file to 05.4_combined_set_with_stereochemistry.tsv.xz. - Embedding The embedding engine that was used to generate 3D coordinates. Description of the molecular descriptors available in Papyrus: The 05.5_combined_2D_moldescs_CDDDs.tsv.xz file contains Continuous Data-Driven Descriptors (CDDD; Winter et al. (2019) Chem. Sci. 10(6), 1692-1701, DOI: 10.1039/c8sc04175j) calculated from the 2D molecular structures in 05.5_combined_2D_set_without_stereochemistry.sd.xz. The 05.5_combined_2D_moldescs_ECFP6.tsv.xz file contains the RDKit extended connectivity Morgan fingerprints with radius 3 and 2048 bits calculated from the 2D molecular structures in 05.5_combined_2D_set_without_stereochemistry.sd.xz. The 05.5_combined_2D_moldescs_mold2.tsv.xz file contains the Mold2 descriptors (Hong et al. (2008). J. Chem. Inf. Model. 48(7), 1337-1344, DOI: 10.1021/ci800038f) calculated from the 2D molecular structures in 05.5_combined_2D_set_without_stereochemistry.sd.xz. The 05.5_combined_2D_moldescs_mordred2D.tsv.xz file contains Mordred descriptors (Moriwaki et al. (2018) J. Cheminf., DOI: 10.1186/s13321-018-0258-y) calculated from the 2D molecular structures in 05.5_combined_2D_set_without_stereochemistry.sd.xz. The 05.5_combined_3D_moldescs_E3FP.tsv.xz file contains extended three-dimensional fingerprint (E3FP; Axen et al. (2017) J. Med. Chem. 60(17), 7393-7409, DOI: 10.1021/acs.jmedchem.7b00696) calculated from 3D molecular structures in 05.5_combined_3D_set_with_stereochemistry.xd.xz. The 05.5_combined_3D_moldescs_mordred3D.tsv.xz file contains Mordred descriptors calculated from 3D molecular structures in 05.5_combined_3D_set_with_stereochemistry.xd.xz. Description of the protein descriptors available in Papyrus: The 05.5_combined_prot_embeddings_unirep.tsv.xz file contains UniRep protein descriptors (Alley et al. (2019) Nat. Methods 16(12), 1315-1322, DOI: 10.1038/s41592-019-0598-1) calculated from the protein sequences in 05.5_combined_set_protein_targets.tsv.xz.