Autophagy dark genes: Can we find them with machine learning?

Abstract

Identifying novel genes associated with autophagy (ATG) in man remains an important task for gaining complete understanding on this fundamental physiological process. A machine-learning guided approach can highlight potentially "missing pieces" linking core autophagy genes with understudied, "dark" genes that can help us gain deeper insight into these processes. In this study, we used a set of 103 (out of 288 genes from the Autophagy Database, ATGdb), based on the presence of ATG-associated terms annotated from 3 secondary sources: GO (gene ontology), KEGG pathway and UniProt keywords, respectively. We regarded these as additional confirmation for their importance in ATG. As negative labels, we used the OMIM list of genes associated with monogenic diseases (after excluding the 288 ATG-associated genes). Data associated with these genes from 17 different public sources were compiled and used to derive a Meta Path/XGBoost (MPxgb) machine learning model trained to distinguish ATG and non-ATG genes (10-fold cross-validated, 100-times randomized models, median AUC = 0.994 +/-0.0084). Sixteen ATG-relevant variables explain 64% of the total model gain, and 23% of the top 251 predicted genes are annotated in ATGdb. Another 15 genes have potential ATG associations, whereas 193 do not. We suggest that some of these 193 genes may represent "autophagy dark genes", and argue that machine learning can be used to guide autophagy research in order to gain a more complete functional and pathway annotation of this complex process.

A Workflow of Integrated Resources to Catalyze Network Pharmacology Driven COVID-19 Research

“…The TCRD-KG nodes can be proteins, diseases, or phenotypes, and edges can be pathways, protein–protein interactions, or other biological relationships among proteins and diseases. A machine learning (ML) framework based on the TCRD-KG metapaths and XGBoost classification algorithm was developed to predict disease-associated genes (proteins). The metapaths specify network paths that connect proteins to specific diseases in the TCRD-KG.…”

Autophagy dark genes: Can we find them with machine learning?

“…56 The other two, UFM1 and INPP4A, would not be influenced by "data leakage." 56 In other words, the MPxgb model built on the 2019 KG model 79 did not "leak" ATG-related terms for 2 of the 3 confirmed genes from the lowest ranking set.…”

“…To validate MPxgb predictions, we explored the top 20 predicted genes for literature reports related to ATG because we suggested these as "ATG dark genes" in 2019. 79 All the gene aliases were obtained from the NCBI Gene database 95 to ensure query completeness. The literature search was conducted in PubMed (https://pubmed.ncbi.nlm.nih.…”

A critical overview of computational approaches employed for COVID-19 drug discovery

“…VHPPI sources included two proteomic studies, 14,127 the SARS-CoV-2 subset from the viral-human interactions atlas 163 and a genome-wide CRISPR screen for host genes related to SARS-CoV-2 infection. 164 To streamline these non-overlapping VHPPIs with hPPIs, 14,127,164 the authors used a KG based machine learning step (described elsewhere in the context of autophagy), 165 by using the “positive” (known) interactions against true negatives (from the above experiments) in the context of data aggregated from 17 distinct machine-learning ready sets from TCRD/Pharos. 166 For the pharmacology component of the network, DTIs were extracted from the DrugCentral database.…”