In the environment, chemicals remain in a mixture form, whereas in traditional QSAR, models are developed usually for isolated compounds. Chemicals in a mixture form may behave in a different way due to the interactions with and effects of other chemicals. The lack of reliable data poses one of the biggest challenges for the development of QSARs for toxicity of chemical mixtures. In addition, proper external validation is less straightforward for QSAR models of mixtures because the same compounds with different ratios may be present in the data set. The QSAR modeling of mixtures requires the use of appropriate descriptors. Some of the approaches already applied [7] include descriptors based on the mixture partition coefficient [8], additive molecular descriptors [9], integral nonadditive descriptors of mixtures [10], and fragment nonadditive descriptors [11]. Modeling of mixture toxicity is a relatively new application field of QSAR; further efforts are to be directed to the development of new methods (including descriptors) and the improvement of existing QSAR approaches for mixtures.

In view of increasing antibiotic resistance by pathogens, antimicrobial peptides have drawn significant attention as an alternative class of antimicrobial therapeutics. Although a broad spectrum of antimicrobial peptides have been reported, their structure–activity relationships (SAR) are not well understood, largely because of substantial diversity in their structures and their nonspecific mechanism of action. There is the possibility that QSAR could be applied in further understanding their SAR. The majority of the previous work in this direction was sequence-based modeling efforts in a qualitative manner. There are some recent reports [12–15] on residue- and atom-based approaches in modeling antimicrobial peptides. It has emerged [7] that the atomic level of consideration combined with machine learning techniques may result in models delivering more active peptides. More extensive studies are needed in this area.

Nanoparticles (NPs) have found a wide range of applications in different fields of human life. They are employed in diverse industrial sectors, such as electronics, biomedical, pharmaceutical, cosmetics, and many others. Interestingly, our understanding regarding the harmful interactions of NPs with biological systems, as well as with the environment, is insufficient, and also the current understanding of the toxicity of NP, including possible mutagenic and carcinogenic effects, is very limited. In this background, theoretical methods like QSAR modeling might be applicable for the comprehensive risk exposure and assessment of NPs at the early stage of their development [16]. Information on the properties and toxicities of NPs is required under multiple regulatory documents, including REACH, the Biocides Directive, the Plant Protection Products Directive, the Water Framework Directive, and the Cosmetics Directive, where information shall be generated whenever possible by means other than vertebrate animal tests, through the use of alternative methods such as in vitro methods or QSAR models. The term nano-QSAR has recently been coined to describe this. A few nano-QSAR models have been recently reported for different end points, including cellular uptake of NPs in pancreatic cancer cell (PaCa2) [17–19], toxicity of metal oxide NPs for human bronchial epithelial (BEAS-2B) and murine myeloid (RAW 264.7) [20], and for the nanotoxicity of surface-modified multiwalled carbon nanotubes [21].

The most important encountered problems for developing QSAR models of NPs are the following: