Dataset Information

A deep learning framework for the reconstruction of gene expression profiles derived from formalin-fixed paraffin-embedded samples

ABSTRACT: This dataset was applied to evaluate the performance of a deep learning framework, FFPERescuer, specifically designed to reconstruct gene expression profiles from RNA sequencing data derived from FFPE (formalin-fixed, paraffin-embedded) tumor tissues. The dataset includes a total of 12 RNA-Seq samples, comprising 10 FFPE tumor tissue samples from colorectal cancer (CRC) cases collected in Amsterdam, the Netherlands, with 2 duplicate samples included for reproducibility assessment. The corresponding gene expression profiles from fresh-frozen tumor tissues for these 10 cases are available in the dataset GSE33113, generated using microarray technology.

ORGANISM(S): Homo sapiens

PROVIDER: GSE299799 | GEO | 2025/07/31

REPOSITORIES: GEO

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:For the effectiveness of targeted immunotherapy, which is currently one of the most promising treatments for brain tumors, it is necessary to know specific tumor-associated antigens (TAA). This project aimed to validate the suitability of formalin-fixed and paraffin-embedded (FFPE) material instead of fresh-frozen material for RNA sequencing and downstream TAA identification. Purpose: Detect potential oligodendroglioma-associated antigens using both fresh-frozen and FFPE brain tissues; assess the suitability of the FFPE material for TAA identification. Results: A comparative analysis based on the RNA-seq of canine oligodendroglioma showed that formalin fixation of the samples had a significant effect on the RNA-seq library in terms of quality. Read distribution analysis showed that the FFPE samples contained fewer reads mapping to exonic regions and were enriched with reads mapped to introns and intergenic regions, while the fresh-frozen samples were mostly enriched in reads mapping to exons. Mismatch profile and SNVs calling analyses revealed some substitution artefacts present in the FFPE samples and absent in the corresponding fresh-frozen samples. However, according to the Principal Component Analysis, 36% of the variance between the samples could be explained by the types of conservation, while 43% of the variance could still be attributed to different expression profiles between oligodendroglioma and control conditions. A differential expression analysis of fresh-frozen oligodendroglioma versus fresh-frozen control samples identified 62 potential TAA strongly up-regulated in tumor tissue. The same analysis using the FFPE samples showed 80% of these genes (49/62) also to be differentially expressed. Comparative analysis showed good agreement between FFPE and fresh-frozen RNA-seq libraries in terms of the accuracy of gene expression measurements indicating that archived FFPE tissue can be used for identification of TAA candidates by RNA-seq providing a wealthy source of clinical samples for research. The following 10 genes encoding cell surface proteins have been identified by both approaches as potential TAA candidates based on their strong overexpression in oligodendrogliomas: PDGFRA, NOTCH1, DLL1, GPER1, TNR, IQGAP3, CD44, ERBB3, BCAS1 and ROR2. Future projects still need to investigate their suitability as TAA for targeted immunotherapy. Conclusion: Formalin fixation of the samples had a significant impact on the RNA-seq quality in terms of the accuracy of gene expression measurements, as well as the quality of the nucleotide sequences. Nevertheless, the comparative analysis showed good agreement between FFPE and fresh-frozen RNA-seq libraries.

Dataset Information

A deep learning framework for the reconstruction of gene expression profiles derived from formalin-fixed paraffin-embedded samples

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets