As the application of large language models (LLMs) has broadened into the realm of biological predictions, leveraging their capacity for self-supervised learning to create feature representations of amino acid sequences, these models have set a new benchmark in tackling downstream challenges, such as subcellular localization. However, previous studies have primarily focused on either the structural design of models or differing strategies for fine-tuning, largely overlooking investigations into the nature of the features derived from LLMs. In this research, we propose different ESM2 representation extraction strategies, considering both the character type and position within the ESM2 input sequence. Using model dimensionality reduction, predictive analysis and interpretability techniques, we have illuminated potential associations between diverse feature types and specific subcellular localizations. Particularly, the prediction of Mitochondrion and Golgi apparatus prefer segments feature closer to the N-terminal, and phosphorylation site-based features could mirror phosphorylation properties. We also evaluate the prediction performance and interpretability robustness of Random Forest and Deep Neural Networks with varied feature inputs. This work offers novel insights into maximizing LLMs’ utility, understanding their mechanisms, and extracting biological domain knowledge. Furthermore, we have made the code, feature extraction API, and all relevant materials available at https://github.com/yujuan-zhang/feature-representation-for-LLMs.

Therapeutic effects of anticancer medicines can be improved by targeting the specific receptors on cancer cells. Folate receptor (FR) targeting with antibody (Ab) is an effective tool to deliver anticancer drugs to the cancer cell. In this research project, a novel formulation of targeting drug delivery was designed, and its anticancer effects were analyzed. Folic acid-conjugated magnetic nanoparticles (MNPs) were used for the purification of folate receptors through a novel magnetic affinity purification method. Antibodies against the folate receptors and methotrexate (MTX) were developed and characterized with enzyme-linked immunosorbent assay and Western blot. Targeting nanomedicines (MNP-MTX-FR Ab) were synthesized by engineering the MNP with methotrexate and anti-folate receptor antibody (anti-FR Ab). The cytotoxicity of nanomedicines on HeLa cells was analyzed by calculating the % age cell viability. A fluorescent study was performed with HeLa cells and tumor tissue sections to analyze the binding efficacy and intracellular tracking of synthesized nanomedicines. MNP-MTX-FR Ab demonstrated good cytotoxicity along all the nanocomposites, which confirms that the antibody-coated medicine possesses the potential affinity to destroy cancer cells in the targeted drug delivery process. Immunohistochemical approaches and fluorescent study further confirmed their uptake by FRs on the tumor cells’ surface in antibody-mediated endocytosis. The current approach is a useful addition to targeted drug delivery for better management of cancer therapy along with immunotherapy in the future.

Pyrazinamide (PZA) is the only drug for the elimination of latent Mycobacterium tuberculosis (MTB) isolates. However, due to the increased number of PZA-resistance, the chances of the success of global TB elimination seems to be more prolonged. Recently, marine natural products (MNPs) as an anti-TB agent have received much attention, where some compounds extracted from marine sponge, Haliclona sp. exhibited strong activity under aerobic and hypoxic conditions. In this study, we screened articles from 1994 to 2019 related to marine natural products (MNPs) active against latent MTB isolates. The literature was also mined for the major regulators to map them in the form of a pathway under the dormant stage. Five compounds were found to be more suitable that may be applied as an alternative to PZA for the better management of resistance under latent stage. However, the mechanism of actions behind these compounds is largely unknown. Here, we also applied synthetic biology to analyze the major regulatory pathway under latent TB that might be used for the screening of selective inhibitors among marine natural products (MNPs). We identified key regulators of MTB under latent TB through extensive literature mining and mapped them in the form of regulatory pathway, where SigH is negatively regulated by RshA. PknB, RshA, SigH, and RNA polymerase (RNA-pol) are the major regulators involved in MTB survival under latent stage. Further studies are needed to screen MNPs active against the main regulators of dormant MTB isolates. To reduce the PZA resistance burden, understanding the regulatory pathways may help in selective targets of MNPs from marine natural sources.

Anti-cancer peptides (ACPs) play a vital role in the cell signaling process. Antimicrobial peptides (AMPs) provide immunity against pathogenic microbes, AMPs present activity against pathogenic microbes. Some of them are known to possess both anticancer and antimicrobial activity. However, so far, no tools have been developed that could predict potential ACPs from wild and mutated cancerous protein sequences in the numerous public databases. In the present study, we developed a A-CaMP tool that allows rapid fingerprinting of the anti-cancer and antimicrobial peptides, which play a crucial role in current bioinformatics research. Besides, we compared the performance and functionality of our A-CaMP tool with those of other methods available online. A-CaMP scans the target protein sequences provided by the user against the datasets. It possesses a robust coding architecture, has been developed in PERL language and is scalable of therefore has extensive applications in bioinformatics. It was observed to achieve a prediction accuracy of 93.4%, which is much higher than that of any of the existing tools. Sequence alignment studies also highlight the potential use of A-CaMP as a tool for the identification of AMPs. A-CaMP is the first open source tool that uses clinical data and proposes final peptides along with the necessary information; this includes wild and mutant sequence and peptides, which lays the foundation for its application in therapies for cancer and bacterial infections.

Eukaryotic cells contain a huge variety of internally specialized subcellular compartments. Stoichiogenomics aims to reveal patterns of elements usage in biological macromolecules. However, the stoichiogenomic characteristics and how they adapt to various subcellular microenvironments are still unknown. Here we first updated the definition of stoichiogenomics. Then we applied it to subcellular research, and detected distinctive nitrogen content of nuclear and hydrogen, sulfur content of extracellular proteomes. Specially, we found that acidic amino acids (AAs) content of cytoskeletal proteins is the highest. The increased charged AAs are mainly caused by the eukaryotic originated cytoskeletal proteins. Functional subdivision of the cytoskeleton showed that activation, binding/association, and complexes are the three largest functional categories. Electrostatic interaction analysis showed an increased electrostatic interaction between both primary sequences and PPI interfaces of 3D structures, in the cytoskeleton. This study creates a blueprint of subcellular stoichiogenomic characteristics, and explains that charged AAs of the cytoskeleton increased greatly in evolution, which offer material basis for the eukaryotic cytoskeletal proteins to act in two ways of electrostatic interactions, and further perform their activation, binding/association and complex formation.