Interpreting biologically informed neural networks for enhanced proteomic biomarker discovery and pathway analysis
Ontology highlight
ABSTRACT:
INSTRUMENT(S): Q Exactive HF
ORGANISM(S): Homo Sapiens (human)
TISSUE(S): Blood Plasma
SUBMITTER: 
Erik Hartman
LAB HEAD: Johan Malmström
PROVIDER: PXD044264 | Pride | 2023-09-05
REPOSITORIES: Pride
Dataset's files
| Action | DRS | |||
|---|---|---|---|---|
| CK_P1912_146.raw | Raw | |||
| CK_P1912_147.raw | Raw | |||
| CK_P1912_148.raw | Raw | |||
| CK_P1912_150.raw | Raw | |||
| CK_P1912_151.raw | Raw |
Items per page: 5 1 - 5 of 59 |
Similar Datasets
Project description:The statistical validation of peptide and protein identifications in mass spectrometry proteomics is a critical step in the analytical workflow. This is particularly important in discovery experiments to ensure only confident identifications are accumulated for downstream analysis and biomarker consideration. However, the inherent nature of discovery proteomics experiments leads to scenarios where the search space will inflate substantially due to the increased number of potential proteins that are being queried in each sample. In these cases, issues will begin to arise when the machine learning algorithms that are trained on an experiment specific basis cannot accurately distinguish between correct and incorrect identifications and will struggle to accurately control the false discovery rate. Here, we propose an alternative validation algorithm trained on a curated external data set of 2.8 million extracted peakgroups that leverages advanced machine learning techniques to create a generalizable peakgroup scoring (GPS) method for data independent acquisition (DIA) mass spectrometry. By breaking the reliance on the experimental data at hand and instead training on a curated external dataset, GPS can confidently control the false discovery rate while increasing the number of identifications and providing more accurate quantification in different search space scenarios. To first test the performance of GPS in a standard experimental environment and to provide a benchmark against other methods, a novel spike-in data set with known varying concentrations was analyzed. When compared to existing methods GPS increased the nunmber of identifications by 5-18% and was able to provide more accurate quantification by increasing the number of ratio validated identifications by 24-74%. To evaluate GPS in a larger search space, a novel data set of 141 blood plasma samples from patients developing acute kidney injury after sepsis was searched with a human tissue spectral library (10000+ proteins). Using GPS, we were able to provide a 207-377% increase in the number of candidate differentially abundant proteins compared to the existing methods while maintaining competitive numbers of global identifications. Finally, using an optimized human tissue library and workflow we were able to identify 1205 proteins from the 141 plasma samples and increase the number of candidate differentially abundant proteins by 70.87%. With the addition of machine learning aided differential expression, we were able to identify potential new biomarkers for stratifying subphenotypes of acute kidney injury in sepsis. These findings suggest that by using a generalized model such as GPS in tandem with a massive scale spectral library it is possible to expand the boundaries of discovery experiments in DIA proteomics. GPS is open source and freely available on github at https://github.com/InfectionMedicineProteomics/gps
2023-07-20 | PXD038394 | Pride
Project description:The statistical validation of peptide and protein identifications in mass spectrometry proteomics is a critical step in the analytical workflow. This is particularly important in discovery experiments to ensure only confident identifications are accumulated for downstream analysis and biomarker consideration. However, the inherent nature of discovery proteomics experiments leads to scenarios where the search space will inflate substantially due to the increased number of potential proteins that are being queried in each sample. In these cases, issues will begin to arise when the machine learning algorithms that are trained on an experiment specific basis cannot accurately distinguish between correct and incorrect identifications and will struggle to accurately control the false discovery rate. Here, we propose an alternative validation algorithm trained on a curated external data set of 2.8 million extracted peakgroups that leverages advanced machine learning techniques to create a generalizable peakgroup scoring (GPS) method for data independent acquisition (DIA) mass spectrometry. By breaking the reliance on the experimental data at hand and instead training on a curated external dataset, GPS can confidently control the false discovery rate while increasing the number of identifications and providing more accurate quantification in different search space scenarios. To first test the performance of GPS in a standard experimental environment and to provide a benchmark against other methods, a novel spike-in data set with known varying concentrations was analyzed. When compared to existing methods GPS increased the nunmber of identifications by 5-18\% and was able to provide more accurate quantification by increasing the number of ratio validated identifications by 24-74\%. To evaluate GPS in a larger search space, a novel data set of 141 blood plasma samples from patients developing acute kidney injury after sepsis was searched with a human tissue spectral library (10000+ proteins). Using GPS, we were able to provide a 207-377\% increase in the number of candidate differentially abundant proteins compared to the existing methods while maintaining competitive numbers of global identifications. Finally, using an optimized human tissue library and workflow we were able to identify 1205 proteins from the 141 plasma samples and increase the number of candidate differentially abundant proteins by 70.87\%. With the addition of machine learning aided differential expression, we were able to identify potential new biomarkers for stratifying subphenotypes of acute kidney injury in sepsis. These findings suggest that by using a generalized model such as GPS in tandem with a massive scale spectral library it is possible to expand the boundaries of discovery experiments in DIA proteomics. GPS is open source and freely available on github at (\url{https://github.com/InfectionMedicineProteomics/gps})
2023-07-20 | PXD038377 | Pride
Project description:Interstitial lung diseases such as idiopathic pulmonary fibrosis (IPF) are caused by persistent micro-injuries to alveolar epithelial tissues accompanied by aberrant repair processes. Despite substantial advancement in our understanding of IPF progression, numerous questions remain concerning disease pathology. IPF is currently treated with pirfenidone and nintedanib, compounds which slow the rate of disease progression but fail to target underlying pathophysiological mechanisms. The DNA repair enzyme 8-oxoguanine DNA glycosylase-1 (OGG1) is upregulated following TGF-β1 exposure in several fibrosis-associated cell types. Currently, no pharmaceutical solutions targeting OGG1 have been utilized in the treatment of IPF. In this study, administration of Ogg1-targetting siRNA, mitigated bleomycin-induced pulmonary fibrosis in mice, thereby highlighting OGG1 as a tractable target in lung fibrosis. The novel small molecule OGG1 inhibitor, TH5487, decreased myofibroblast transition and associated pro-fibrotic markers in fibroblast cells. In addition, TH5487 decreased pro-inflammatory cytokine production, inflammatory cell infiltration, and lung remodeling in a murine model of bleomycin-induced pulmonary fibrosis. OGG1 and SMAD7 interact to induce fibroblast proliferation and differentiation, with both increased in fibrotic murine and IPF patient lung tissue. Taken together, these data strongly suggest that TH5487 is a potent, specific, and clinically-relevant treatment for IPF. This DIA-MS dataset entails the raw data and peptide-centric DIA-NN search results of both, lung tissue and bronchoalveolar lavage fluid of n=5 mice profiled across the treatment groups bleomycin combined with TH (BTH), dexamethasone (DEX), TH alone (TH) and vehicle control (V) relative to bleomycin alone (B) as control. Different animals within protein groups were considered biological replicates of the respective treatment condition.
2023-02-06 | PXD029625 | Pride
Project description:Understanding the plasma proteome dynamics in sepsis patients is crucial for improving prognostic and therapeutic strategies, as sepsis remains a leading cause of mortality in intensive care units. In this study, 363 patients with newly diagnosed sepsis were enrolled and plasma was collected on the first and fourth day after diagnosis. Proteome analysis was performed using liquid chromatography–tandem mass spectrometry. The plasma proteome was analyzed via classical statistical analysis and machine learning to identify associations with 30-day survival. Out of 363 sepsis patients, 224 survived and 139 did not survive the 30-day period. Significant differences in the abundance of 87 proteins on day 1 and 95 proteins on day 4 were found between survivors and non-survivors. Additionally, between days 1 and 4, 63 proteins were differentially regulated between both groups. We found that protein regulations within the first days of sepsis were generally associated with a worse outcome. Statistical analysis revealed the underlying proteins to be foremost related to blood coagulation, immune response and complement activation. Complementarily, the machine learning classifier achieved an area under the receiver operating characteristic curve of 0.75 for predicting 30-day survival. The feature importance analysis highlighted additional proteins and substantiated the findings of univariate statistics. This study describes the temporal alterations of the plasma proteome and the underlying protein networks that are associated with survival. These findings enhance the understanding of sepsis pathophysiology and the identified proteins might serve as starting points for further investigations and guide the development of targeted therapeutic strategies.
2024-12-16 | PXD055932 | Pride
Project description:CpG DNA binds to Toll-like receptor 9 to stimulate a strong innate immune response. Despite extensive studies of TLR9 mediated signal transduction, the scope of CpG-induced changes in gene expression is incompletely understood. In particular, the prolonged effects of CpG ODN (oligodeoxynucleotide) on gene activation have not been investigated despite evidence that a single dose of CpG ODN alters immune reactivity for several weeks. This study used gene expression analysis to monitor changes in mRNA levels for 14 days, and identified the genes, pathways, functional groups and regulatory networks triggered in vivo following CpG ODN administration. Two discrete peaks of gene activation (at 3 hr and 5 days) were observed after CpG administration. Both the regulation and function of genes activated during the second peak differed from those triggered shortly after CpG administration. Initial gene up-regulation corresponded to a period when TLR9 ligation stimulated genes functionally associated with the generation of innate and adaptive immune responses (e.g. the NF-kB and B-cell receptor pathways). The second peak reflected processes associated with cell division (e.g., cell cycle and DNA replication & repair). Whereas TNF and IFNG played central roles regulating the first peak of activation, E2F1, E2F2, BRCA1, and HRAS had major impact on the networks controlling the second peak. The complex bimodal pattern of gene expression elicited by CpG ODN administration provides novel insights into the long term effects of CpG DNA on genes associated with immunity and cell proliferation. Data from 4 independent biological replicates/time point (9 time points) and 6 untreated controls were used for all statistical analyses. A reference design was used. Reference cDNA vs. sample cDNA were hybridized to same array.
2010-05-16 | E-GEOD-16465 | biostudies-arrayexpress
Project description:Burn wound blister fluid is a valuable matrix for understanding the biological pathways associated with burn injury. In this study, 152 blister fluid samples collected from paediatric burn wounds at two different hospitals were analysed using mass spectrometry proteomic techniques. The protein abundance profile at different days post-burn indicated that there were more proteins associated with cellular damage/repair in the first 24 hours, whereas after this point there were more proteins associated with antimicrobial defence and inflammation. The inflammatory proteins persisted at a high level in the blister fluid for more than 7 days. This may indicate that removal of burn blisters prior to two days post-burn is optimal to prevent excessive or prolonged inflammation in the wound environment. Additionally, many proteins associated with the neutrophil extracellular trap (NET) pathway were increased post-burn, further implicating NETs in the post-burn inflammatory response. NET inhibitors may therefore be a potential treatment to reduce post-burn inflammation and coagulation pathology and enhance burn wound healing outcomes.
2024-03-04 | PXD050003 | Pride
Project description:The onset of an infection-specific transcriptional program precedes the clinical diagnosis in patients who developed Ventilator-associated pneumonia (VAP). Ventilator-associated tracheobronchitis (VAT) is another respiratory infection affecting<br><br>outcomes in intubated patients, but interactions between VAT and VAP remains unknown.
2011-11-19 | E-MEXP-3001 | biostudies-arrayexpress
Project description:The role of Krüppel-associated box zinc finger proteins (KZFPs), the largest family of repressive transcription factors (TFs) of the human genome, in the pathogenesis of diffuse large B cell lymphoma has not been addressed so far. By correlating the expression of KZFPs with patient outcome, we delineated a subset of 18 co-expressed KZFPs associated to poor prognosis, higher copy number alterations (CNA), and the suppression of hallmarks gene sets linked to immune rejection. Among them, the depletion of the ZNF587-ZNF417 pair of TF paralogs drastically impaired the proliferation of both activated B cell (ABC) and germinal center B (GCB) DLBCL cell lines, by triggering replicative stress through the awakening of dormant replication origins and profound alterations of the heterochromatin landscape. ZNF587/417 knockdown profoundly altered the transcriptomic landscape of U2932 (ABC) and OCI-Ly7 (GCB) cells, rewiring gene expression networks associated with good prognosis and upregulating interferon (IFN)/inflammatory-related genes through activation of the cGAS-STING DNA sensing pathway. These cells showed signs of altered immunogenicity with an enhanced MHC class I surface expression and a reshuffling of their immunopeptidome. Our study provides evidence that the ZNF587-ZNF417 paralog pair may promote lymphoma growth, the acquisition of genetic diversity, and its immune evasion by preventing excessive replicative stress. This suggests that these TFs participate in the epigenetic memory of lymphoma cells, thus highlighting the critical role of constitutive heterochromatin in maintaining the identity of poor-prognosis DLBCL.
2024-03-15 | PXD040500 | Pride
Project description:Mesenchymal stromal stem cells (MSCs) gain increasing focus in the field of regenerative medicine due to their differentiational abilities into chondrocytes, adipocytes and osteoblastic cells. However, it is apparent that the transformation processes are extremely complex, cause cellular heterogeneity and need further insights into functional annotation. The aim of the study was to characterize differences between MSCs and cells after adipogenic or osteoblastic differentiation at the proteome level. For this, we analysed the proteome of isolated and cultured MSCs and after differentiation into adipogenic and osteoblastic cell lineages using electrospray ionization tandem mass spectrometry in data-independent acquisition mode. Protein inference and quantification was performed using deep neural networks in library free mode. By jointly using strict differential expression analysis and machine learning algorithms, two cell type specific marker protein panels for the osteoblastic and the adipocytic lineage were identified which allow for distinguishment of the three cell types.
2022-03-25 | PXD029900 | Pride
Project description:The prevalence of Parkinson's disease (PD) is increasing but the development of novel treatment strategies and therapeutics altering the course of the disease would benefit from specific, sensitive and non-invasive biomarkers to detect PD early. Here, we describe a scalable and sensitive proteomics workflow for urinary proteome profiling by combining high-throughput sample preparation with state-of-the-art mass spectrometry (MS)-based proteomics. Our workflow enabled the reproducible quantification of more than 2,000 proteins in more than 200 urine samples using minimal volumes from two independent cohorts. The urinary proteome was significantly different between PD patients and healthy controls as well as between LRRK2 G2019S carriers and non-carriers in both cohorts. Interestingly, our data revealed lysosomal dysregulation in individuals with the LRRK2 G2019S mutation. Machine learning on the urinary proteome data alone classified mutation status and especially disease manifestation in mutation carriers remarkably well (ROC AUCs 0.87 and 0.94, respectively), identifying VGF, ENPEP and other PD-associated proteins as the most discriminating features. Our results validate urinary proteomics as a valuable strategy for biomarker discovery and patient stratification in PD.
2020-11-24 | PXD020722 | Pride