Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models
Drug resistance to anticancer drugs is a serious complication in patients with cancer. Typically, drug resistance occurs due to amino acid substitutions (AAS) in drug target proteins. The study aimed at developing and validating a new approach to the creation of structure-property relationships (SPR...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-08-01
|
| Series: | Life |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-1729/13/9/1807 |
| _version_ | 1797579170286403584 |
|---|---|
| author | Svetlana I. Zhuravleva Anton D. Zadorozhny Boris V. Shilov Alexey A. Lagunin |
| author_facet | Svetlana I. Zhuravleva Anton D. Zadorozhny Boris V. Shilov Alexey A. Lagunin |
| author_sort | Svetlana I. Zhuravleva |
| collection | DOAJ |
| description | Drug resistance to anticancer drugs is a serious complication in patients with cancer. Typically, drug resistance occurs due to amino acid substitutions (AAS) in drug target proteins. The study aimed at developing and validating a new approach to the creation of structure-property relationships (SPR) classification models to predict AASs leading to drug resistance to inhibitors of tyrosine-protein kinase ABL1. The approach was based on the representation of AASs as peptides described in terms of structural formulas. The data on drug-resistant and non-resistant variants of AAS for two isoforms of ABL1 were extracted from the COSMIC database. The given training sets (approximately 700 missense variants) were used for the creation of SPR models in MultiPASS software based on substructural atom-centric multiple neighborhoods of atom (MNA) descriptors for the description of the structural formula of protein fragments and a Bayesian-like algorithm for revealing structure-property relationships. It was found that MNA descriptors of the 6th level and peptides from 11 amino acid residues were the best combination for ABL1 isoform 1 with the prediction accuracy (AUC) of resistance to imatinib (0.897) and dasatinib (0.996). For ABL1 isoform 2 (resistance to imatinib), the best combination was MNA descriptors of the 6th level, peptides form 15 amino acids (AUC value was 0.909). The prediction of possible drug-resistant AASs was made for dbSNP and gnomAD data. The six selected most probable imatinib-resistant AASs were additionally validated by molecular modeling and docking, which confirmed the possibility of resistance for the E334V and T392I variants. |
| first_indexed | 2024-03-10T22:33:05Z |
| format | Article |
| id | doaj.art-7928706d4fe0491fb310e9323029a73d |
| institution | Directory Open Access Journal |
| issn | 2075-1729 |
| language | English |
| last_indexed | 2024-03-10T22:33:05Z |
| publishDate | 2023-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Life |
| spelling | doaj.art-7928706d4fe0491fb310e9323029a73d2023-11-19T11:36:34ZengMDPI AGLife2075-17292023-08-01139180710.3390/life13091807Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification ModelsSvetlana I. Zhuravleva0Anton D. Zadorozhny1Boris V. Shilov2Alexey A. Lagunin3Department of Bioinformatics, Pirogov Russian National Research Medical University, 117997 Moscow, RussiaDepartment of Bioinformatics, Pirogov Russian National Research Medical University, 117997 Moscow, RussiaDepartment of Bioinformatics, Pirogov Russian National Research Medical University, 117997 Moscow, RussiaDepartment of Bioinformatics, Pirogov Russian National Research Medical University, 117997 Moscow, RussiaDrug resistance to anticancer drugs is a serious complication in patients with cancer. Typically, drug resistance occurs due to amino acid substitutions (AAS) in drug target proteins. The study aimed at developing and validating a new approach to the creation of structure-property relationships (SPR) classification models to predict AASs leading to drug resistance to inhibitors of tyrosine-protein kinase ABL1. The approach was based on the representation of AASs as peptides described in terms of structural formulas. The data on drug-resistant and non-resistant variants of AAS for two isoforms of ABL1 were extracted from the COSMIC database. The given training sets (approximately 700 missense variants) were used for the creation of SPR models in MultiPASS software based on substructural atom-centric multiple neighborhoods of atom (MNA) descriptors for the description of the structural formula of protein fragments and a Bayesian-like algorithm for revealing structure-property relationships. It was found that MNA descriptors of the 6th level and peptides from 11 amino acid residues were the best combination for ABL1 isoform 1 with the prediction accuracy (AUC) of resistance to imatinib (0.897) and dasatinib (0.996). For ABL1 isoform 2 (resistance to imatinib), the best combination was MNA descriptors of the 6th level, peptides form 15 amino acids (AUC value was 0.909). The prediction of possible drug-resistant AASs was made for dbSNP and gnomAD data. The six selected most probable imatinib-resistant AASs were additionally validated by molecular modeling and docking, which confirmed the possibility of resistance for the E334V and T392I variants.https://www.mdpi.com/2075-1729/13/9/1807sequence-structure analysisdrug resistancemolecular fragmentsMNA descriptorsMultiPASSamino acid substitution |
| spellingShingle | Svetlana I. Zhuravleva Anton D. Zadorozhny Boris V. Shilov Alexey A. Lagunin Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models Life sequence-structure analysis drug resistance molecular fragments MNA descriptors MultiPASS amino acid substitution |
| title | Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models |
| title_full | Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models |
| title_fullStr | Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models |
| title_full_unstemmed | Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models |
| title_short | Prediction of Amino Acid Substitutions in ABL1 Protein Leading to Tumor Drug Resistance Based on “Structure-Property” Relationship Classification Models |
| title_sort | prediction of amino acid substitutions in abl1 protein leading to tumor drug resistance based on structure property relationship classification models |
| topic | sequence-structure analysis drug resistance molecular fragments MNA descriptors MultiPASS amino acid substitution |
| url | https://www.mdpi.com/2075-1729/13/9/1807 |
| work_keys_str_mv | AT svetlanaizhuravleva predictionofaminoacidsubstitutionsinabl1proteinleadingtotumordrugresistancebasedonstructurepropertyrelationshipclassificationmodels AT antondzadorozhny predictionofaminoacidsubstitutionsinabl1proteinleadingtotumordrugresistancebasedonstructurepropertyrelationshipclassificationmodels AT borisvshilov predictionofaminoacidsubstitutionsinabl1proteinleadingtotumordrugresistancebasedonstructurepropertyrelationshipclassificationmodels AT alexeyalagunin predictionofaminoacidsubstitutionsinabl1proteinleadingtotumordrugresistancebasedonstructurepropertyrelationshipclassificationmodels |