Methodology for Good Machine Learning with Multi‐Omics Data

Coroller, Thibaud (ORCID:0000000176628724); Sahiner, Berkman (ORCID:0000000328042264); Amatya, Anup; Gossmann, Alexej (ORCID:0000000190683877); Karagiannis, Konstantinos; Moloney, Conor; Samala, Ravi (ORCID:0000000266614801); Santana‐Quintero, Luis (ORCID:0000000242901021); Solovieff, Nadia; Wang, Craig (ORCID:0000000318042463); Amiri‐Kordestani, Laleh; Cao, Qian; Cha, Kenny (ORCID:0000000338477448); Charlab, Rosane; Cross, Frank; Hu, Tingting; Huang, Ruihao (ORCID:0000000158064961); Kraft, Jeffrey; Krusche, Peter; Li, Yutong; Li, Zheng; Mazo, Ilya (ORCID:0000000285874895); Paul, Rahul (ORCID:0000000314911166); Schnakenberg, Susan; Serra, Paolo; Smith, Sean (ORCID:0000000167584037); Song, Chi (ORCID:0000000311081750); Su, Fei (ORCID:0009000341210944); Tiwari, Mohit; Vechery, Colin; Xiong, Xin; Zarate, Juan; Zhu, Hao; Chakravartty, Arunava; Liu, Qi (ORCID:0000000240534213); Ohlssen, David; Petrick, Nicholas (ORCID:0000000151678899); Schneider, Julie (ORCID:0000000307565007); Walderhaug, Mark (ORCID:0000000320392040); Zuber, Emmanuel (ORCID:0000000270312400)

doi:10.1002/cpt.3105

DOE OSTI · 2580605

Methodology for Good Machine Learning with Multi‐Omics Data

Coroller, Thibaud (ORCID:0000000176628724)·Sahiner, Berkman (ORCID:0000000328042264)·Amatya, Anup·Gossmann, Alexej (ORCID:0000000190683877)·Karagiannis, Konstantinos·Moloney, Conor·Samala, Ravi (ORCID:0000000266614801)·Santana‐Quintero, Luis (ORCID:0000000242901021)·Solovieff, Nadia·Wang, Craig (ORCID:0000000318042463)·Amiri‐Kordestani, Laleh·Cao, Qian·Cha, Kenny (ORCID:0000000338477448)·Charlab, Rosane·Cross, Frank·Hu, Tingting·Huang, Ruihao (ORCID:0000000158064961)·Kraft, Jeffrey·Krusche, Peter·Li, Yutong·Li, Zheng·Mazo, Ilya (ORCID:0000000285874895)·Paul, Rahul (ORCID:0000000314911166)·Schnakenberg, Susan·Serra, Paolo·Smith, Sean (ORCID:0000000167584037)·Song, Chi (ORCID:0000000311081750)·Su, Fei (ORCID:0009000341210944)·Tiwari, Mohit·Vechery, Colin·Xiong, Xin·Zarate, Juan·Zhu, Hao·Chakravartty, Arunava·Liu, Qi (ORCID:0000000240534213)·Ohlssen, David·Petrick, Nicholas (ORCID:0000000151678899)·Schneider, Julie (ORCID:0000000307565007)·Walderhaug, Mark (ORCID:0000000320392040)·Zuber, Emmanuel (ORCID:0000000270312400)

View original source ↗Export metadata ↓

Abstract

In 2020, Novartis Pharmaceuticals Corporation and the U.S. Food and Drug Administration (FDA) started a 4‐year scientific collaboration to approach complex new data modalities and advanced analytics. The scientific question was to find novel radio‐genomics‐based prognostic and predictive factors for HR+/HER− metastatic breast cancer under a Research Collaboration Agreement. This collaboration has been providing valuable insights to help successfully implement future scientific projects, particularly using artificial intelligence and machine learning. This tutorial aims to provide tangible guidelines for a multi‐omics project that includes multidisciplinary expert teams, spanning across different institutions. We cover key ideas, such as “maintaining effective communication” and “following good data science practices,” followed by the four steps of exploratory projects, namely (1) plan, (2) design, (3) develop, and (4) disseminate. We break each step into smaller concepts with strategies for implementation and provide illustrations from our collaboration to further give the readers actionable guidance.

Explore related subjects

Pharmacology & Pharmacy

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Coroller, Thibaud (ORCID:0000000176628724), Sahiner, Berkman (ORCID:0000000328042264), Amatya, Anup, Gossmann, Alexej (ORCID:0000000190683877), Karagiannis, Konstantinos, Moloney, Conor, Samala, Ravi (ORCID:0000000266614801), Santana‐Quintero, Luis (ORCID:0000000242901021), Solovieff, Nadia, Wang, Craig (ORCID:0000000318042463), Amiri‐Kordestani, Laleh, Cao, Qian, Cha, Kenny (ORCID:0000000338477448), Charlab, Rosane, Cross, Frank, Hu, Tingting, Huang, Ruihao (ORCID:0000000158064961), Kraft, Jeffrey, Krusche, Peter, Li, Yutong, Li, Zheng, Mazo, Ilya (ORCID:0000000285874895), Paul, Rahul (ORCID:0000000314911166), Schnakenberg, Susan, Serra, Paolo, Smith, Sean (ORCID:0000000167584037), Song, Chi (ORCID:0000000311081750), Su, Fei (ORCID:0009000341210944), Tiwari, Mohit, Vechery, Colin, Xiong, Xin, Zarate, Juan, Zhu, Hao, Chakravartty, Arunava, Liu, Qi (ORCID:0000000240534213), Ohlssen, David, Petrick, Nicholas (ORCID:0000000151678899), Schneider, Julie (ORCID:0000000307565007), Walderhaug, Mark (ORCID:0000000320392040), Zuber, Emmanuel (ORCID:0000000270312400). 2024-01-31. Methodology for Good Machine Learning with Multi‐Omics Data. https://doi.org/10.1002/cpt.3105

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

DOE OSTI2025

Discovery of Macrocyclic Myeloid Cell Leukemia 1 (Mcl-1)Inhibitors that Demonstrate Potent Cellular Efficacy and In Vivo Activity in a Mouse Solid Tumor Xenograft Model

Abstract The B cell lymphoma 2 (Bcl-2) family of proteins are key regulators of intrinsic apoptosis. The antiapoptotic protein myeloid cell leukemia 1 (Mcl-1), which is associated with high tumor grade, poor survival, and resistance to treatment, has emerged as a promising candidate for treating hematological and solid cancers. Herein, we report the structure-guided design of small molecule macrocyclic Mcl-1 inhibitors based on the (R)-methyl-dihydropyrazinoindolone scaffold our group has previously disclosed. The macrocyclic inhibitors bind Mcl-1 with subnanomolar affinity and offer improved potency in cell culture growth inhibition assays. Inhibitor 13 achieved tumor regression in a lung cancer-derived tumor xenograft model in mice as a monotherapy. The improved potency of the macrocyclic series allowed replacement of heretofore conserved indole carboxylic acid moiety, resulting in neutral inhibitors. Amide inhibitor 25 displayed a >10-fold increase in oral bioavailability as compared to acid-containing macrocyclic or acyclic inhibitors.

Pharmacology & Pharmacy↗

DOE OSTI2025

Rationally Designed, Short-Acting RPE65 Inhibitors for Visual Cycle-Associated Retinopathies

Abstract The visual cycle is a metabolic pathway essential for visual function. The bisretinoid byproducts of this pathway can induce retinal toxicity, as occurs in Stargardt disease type 1 (STGD1). Emixustat, which inhibits bisretinoid production, is a visual cycle modulator (VCM) that targets RPE65. However, it causes visual impairment due to its unfavorable duration of action. Here, we report ester-containing analogs of emixustat that are susceptible to hydrolytic clearance and function as short-acting VCMs. We show that the esterase-mediated metabolism of these compounds can be tuned while maintaining high-affinity RPE65 targeting. Compounds 6 (EYE-002) and 7 (EYE-003) containing diethyl acetate and valproate esters, respectively, allowed faster recovery of visual cycle function compared to emixustat. These molecules protected against retinal degeneration in mouse models of photic retinopathy and STGD1. These data demonstrate that shorter attenuation of the visual cycle can therapeutically intervene in retinal diseases with fewer visual side effects compared to emixustat.

Pharmacology & Pharmacy↗

DOE OSTI2025

Identification and Clinical Evaluation of Potential Biomarkers for Breast Cancer Resistance Protein ( BCRP / ABCG2 )

Clinical inhibition and genetic variation of the Breast Cancer Resistance Protein (BCRP/ABCG2) efflux transporter can significantly influence drug exposure, highlighting the need for reliable BCRP functional biomarkers. This study aimed to identify and evaluate biomarkers predictive of BCRP function in humans. A comprehensive analysis of metabolomic genome‐wide association studies (mGWAS) was conducted to discover potential BCRP biomarkers, followed by evaluation inin vitrotransporter assays and a clinical drug–drug interaction (DDI) study. Across multiple mGWAS datasets, plasma concentrations of three herbicide derivatives—4‐hydroxychlorothalonil (4HC), 3‐bromo‐5‐chloro‐2,6‐dihydroxybenzoic acid (BCDBA), and 3,5‐dichloro‐2,6‐dihydroxybenzoic acid (DCDBA)—were significantly elevated (P < 5E‐8) in individuals carrying reduced functionABCG2polymorphisms. These compounds were confirmed as novel BCRP substrates via transporter uptake assays and selected for clinical evaluation alongside riboflavin, a known BCRP substrate and potential BCRP biomarker. In a DDI study with 11 healthy subjects, eltrombopag, a BCRP inhibitor, increased rosuvastatin concentrations by approximately twofold (P = 0.002). No significant changes in the plasma concentrations of organic anion transporting polypeptide 1B (OATP1B) biomarkers (CP‐I and CP‐III) or potential BCRP biomarkers (4HC, BCDBA, DCDBA, or riboflavin) were observed. Notably, two subjects were heterozygous carriers for theABCG2p.Q141K variant and exhibited significantly higher baseline concentrations of 4HC (P = 0.004) and BCDBA (P = 0.0003), consistent with reduced BCRP function. These findings suggest that 4HC and BCDBA are promising biomarkers for baseline BCRP function in specific populations, such as those harboring reduced function genetic polymorphisms, but do not appear suitable for detecting acute BCRP inhibition.

Pharmacology & Pharmacy↗