Discovering and developing new drugs is an expensive, demanding, and time-consuming process with uncertain outcomes in clinical trials [1]. Nowadays, chemical synthesis and biological screening possibilities in the drug discovery pipeline have significantly increased. However, poor pharmacokinetic properties and high toxicity have been considered major limitations of drug candidates. Therefore, lipophilicity assessment is an essential process in drug discovery since it noticeably affects the diffusion of molecules through a biological membrane [2], determining the molecule's pharmacokinetic properties and toxicity [3]. The detailed experimental protocols proposed by OECD (test no 107 and 117) and several procedures developed for academic and industrial institutions describe methods for lipophilicity assessment based on the shake flask method or reversed-phase liquid chromatography (RP-LC) [4], [5], [6], [7], [8], [9], [10]. Nevertheless, currently, more biosimilar alternatives to classical lipophilicity are available. One of the alternatives is immobilized artificial membrane (IAM) chromatography. The IAM stationary phase shows superior biomimetic properties compared to classical lipophilicity measurement since phosphatidylcholine molecules (the primary phospholipid of cell membranes) are covalently bound to silica, mimicking the phospholipid membrane monolayer [11].

The first HPLC columns with an immobilized artificial membrane were developed by Pidgeon et al. in 1989 [12]. IAM high-performance liquid chromatography (IAM-HPLC) allows for assessing the affinity to phospholipids while maintaining all the advantages of HPLC. It is worth emphasizing that chromatographic approaches hold great promise for high-throughput screening among non-cell-based methods. Modern HPLC systems are highly automated and very popular in academia and the pharmaceutical industry [13].

The relationship between retention and the chemical structure of analytes has attracted attention from the beginning of chromatographic research. Kaliszan initiated and introduced a particular type of quantitative structure-property relationships (QSPR) analysis, namely the quantitative structure-retention relationships (QSRR), in 1987 [14]. Since then, QSRR has been a powerful tool in chromatographic research and has also been applied to study the retention mechanism of IAM-HPLC [15], [16], [17], [18], [19], [20], [21].

In the past decades, machine learning (ML) and data sciences have made remarkable progress and are being applied in every scientific discipline, including QSPR/QSRR. Among available algorithms, similarity-based machine learning methods (SBM) showed significant advantages in dealing with heterogeneous noisy toxicological data [22,23]. Nonetheless, the use of SBM in the case of chromatographic data is still in its infancy, although some studies showed its high potential [24].

This study's purpose was to understand better the molecular mechanism of interaction between small xenobiotics and phospholipids. IAM-HPLC data was integrated with molecular descriptors via the QSRR approach to realize this goal. A chromatographic hydrophobicity index with an immobilized artificial membrane (CHIIAM) was used as an endpoint in QSRR modeling. Based on a heterogeneous set of 402 molecules, primarily of pharmaceutical or toxicological significance, the QSRR model was developed. The proposed model was validated using retention factors of potential drug candidates, 106 molecules belonging to 5 different chemical classes, analyzed in our laboratory. Using Chemicalize software, mechanistic molecular descriptors were calculated; therefore, the straightforward interpretation of obtained QSRR models can be described. Initial descriptor selection was performed by applying multiple linear regression (MLR) coupled with a genetic algorithm (GA). Additionally, inspired by recent developments of SBM, especially locally weighted least squares kernel regression (KwLPR), we checked that KwLPR can increase model performances.