Pulmonary fibrosis (PF) is a chronic interstitial lung disease characterized by fibrosis, manifesting clinically as an irksome dry cough, progressive shortness of breath, and reduced lung function (Lynch et al., 2018). Recent studies indicate an increasing incidence of PF, which, due to its significant mortality rate and economic implications, has greatly affected patients' quality of life and emerged as a substantial public health concern (Maher et al., 2021; Spagnolo et al., 2021). However, the precise pathological mechanisms of PF remain unclear, contributing to the limited effective drugs.

The advent of big data has facilitated efficient screening of disease-associated genes, prediction of molecular functions, and research into drug targets and molecular therapies. However, the use of diverse technology platforms and unique clinical samples for individual microarray analyses may results in discrepancies and limitations within the current analytical results. Integrative bioinformatics approaches can address these issues by analyzing multiple datasets, thereby aiding in the screening of multiple disease marker genes (Liu et al., 2019). By integrating differentially expressed genes (DEGs) from holistic biological analyses with key genes identified through WGCNA analysis and refining them using LASSO screening, we can pinpoint signature genes that closely align with the clinical characteristics of the disease. This approach enhances our ability to discover and validate potent and safe drugs for treating PF. Besides, considering the close relationship between PF and immunity (Shenderov et al., 2021), integrating insights from gene targeting with those from immune cell infiltration can deepen our understanding of PF development and pave the way for innovative immunomodulatory therapies Clynick et al. (2022). It emphasizes the importance of identifying more potent targets to unravel the pathophysiological mechanisms of PF Langfelder and Horvath (2008).

Chinese medicine has shown promising efficacy and safety in the treatment of PF, suggesting a vast potential for application. Recently, there has been an increase in studies aiming to elucidate the therapeutic effects and underlying mechanisms of herbal compounds and their active constituents in managing PF(Li and Kan, 2017; Shao et al., 2022; Song et al., 2022). The Platycodon grandiflorum stands out among them as it has demonstrated significant advantages in the treatment of pulmonary diseases. Numerous studies have shown that the saponins of Platycodon grandiflorum are the primary components of its medicinal effects, and with a high level of safety (Shin et al., 2021; Zhang et al., 2022b). Current researches on Platycodon grandiflorum mainly focus on Platycodin D, such as treatment of lung cancer and acute lung injury (ALI), which may be based on its anti-inflammatory and antioxidant activities (Wu et al., 2021; Yim et al., 2016). It is well known that ALI/ARDS without intervention has the potential to eventually transition to the fibrotic stage (Giacomelli et al., 2021). Additionally, the development of PF is closely related to inflammation and oxidative stress (Li et al., 2022). Interestingly, a previous study has shown that both Deapioplatycodin D (DPD) and PD are critical components in the cough suppression of Platycodon grandiflorum, and more importantly, that cough is a symptom that occurs in most patients with early-stage PF (Wakwaya et al., 2021; Zhang et al., 2021). Overall, the presented evidence suggests that DPD, another principle saponin component of Platycodon grandiflorum, may possess anti-PF potential. However, the available evidence regarding the anti-PF effects of DPD is notably scarce. Therefore, further research is required to investigate the effect and underlying molecular mechanism of DPD in PF treatment Strongman et al. (2018).

In this study, lipopolysaccharide (LPS)-induced early PF rat was established to test the overall efficacy of DPD in vivo. To further explore the underlying mechanisms, the GSE53845 and GSE10667 datasets were used as training set to screen DEGs for functional enrichment analysis, and to acquire hub modules highly associated with the development of PF by the WGCNA algorithm, and then used the LASSO regression model to screen key genes and validated them with the GSE24206 dataset. We have not only identified crucial molecular markers, but also assessed the degree of associated immune cell infiltration. Furthermore, LPS-induced BEAS-2B cells was used as a model of early PF, which for validating our predicted results and evaluated the potential of DPD as an anti-PF treatment. This study provides a new basis for the potential molecular mechanisms and therapeutic targets of PF, as well as DPD as a promising anti-PF drug. Flowchart of this study was shown in Graphical Abstract.