Improving gene function predictions using independent transcriptional components

FIRRM and FIGNL1: partners in the regulation of homologous recombination

DNA repair through homologous recombination (HR) is of vital importance for maintaining genome stability and preventing tumorigenesis. RAD51 is the core component of HR, catalyzing the strand invasion and homology search. Here, we highlight recent findings on FIRRM and FIGNL1 as regulators of the dynamics of RAD51.

Independent transcriptional patterns reveal biological processes associated with disease-free survival in early colorectal cancer

BACKGROUND

Bulk transcriptional profiles of early colorectal cancer (CRC) can fail to detect biological processes associated with disease-free survival (DFS) if the transcriptional patterns are subtle and/or obscured by other processes' patterns. Consensus-independent component analysis (c-ICA) can dissect such transcriptomes into statistically independent transcriptional components (TCs), capturing both pronounced and subtle biological processes.

METHODS

In this study we (1) integrated transcriptomes (n = 4228) from multiple early CRC studies, (2) performed c-ICA to define the TC landscape within this integrated data set, 3) determined the biological processes captured by these TCs, (4) performed Cox regression to identify DFS-associated TCs, (5) performed random survival forest (RSF) analyses with activity of DFS-associated TCs as classifiers to identify subgroups of patients, and 6) performed a sensitivity analysis to determine the robustness of our results RESULTS: We identify 191 TCs, 43 of which are associated with DFS, revealing transcriptional diversity among DFS-associated biological processes. A prominent example is the epithelial-mesenchymal transition (EMT), for which we identify an association with nine independent DFS-associated TCs, each with coordinated upregulation or downregulation of various sets of genes.

CONCLUSIONS

This finding indicates that early CRC may have nine distinct routes to achieve EMT, each requiring a specific peri-operative treatment strategy. Finally, we stratify patients into DFS patient subgroups with distinct transcriptional patterns associated with stage 2 and stage 3 CRC.

Exploring combinations of dimensionality reduction, transfer learning, and regularization methods for predicting binary phenotypes with transcriptomic data

BACKGROUND

Numerous transcriptomic-based models have been developed to predict or understand the fundamental mechanisms driving biological phenotypes. However, few models have successfully transitioned into clinical practice due to challenges associated with generalizability and interpretability. To address these issues, researchers have turned to dimensionality reduction methods and have begun implementing transfer learning approaches.

METHODS

In this study, we aimed to determine the optimal combination of dimensionality reduction and regularization methods for predictive modeling. We applied seven dimensionality reduction methods to various datasets, including two supervised methods (linear optimal low-rank projection and low-rank canonical correlation analysis), two unsupervised methods [principal component analysis and consensus independent component analysis (c-ICA)], and three methods [autoencoder (AE), adversarial variational autoencoder, and c-ICA] within a transfer learning framework, trained on > 140,000 transcriptomic profiles. To assess the performance of the different combinations, we used a cross-validation setup encapsulated within a permutation testing framework, analyzing 30 different transcriptomic datasets with binary phenotypes. Furthermore, we included datasets with small sample sizes and phenotypes of varying degrees of predictability, and we employed independent datasets for validation.

RESULTS

Our findings revealed that regularized models without dimensionality reduction achieved the highest predictive performance, challenging the necessity of dimensionality reduction when the primary goal is to achieve optimal predictive performance. However, models using AE and c-ICA with transfer learning for dimensionality reduction showed comparable performance, with enhanced interpretability and robustness of predictors, compared to models using non-dimensionality-reduced data.

CONCLUSION

These findings offer valuable insights into the optimal combination of strategies for enhancing the predictive performance, interpretability, and generalizability of transcriptomic-based models.

Patterns of transcription factor binding and epigenome at promoters allow interpretable predictability of multiple functions of non-coding and coding genes

Understanding the biological roles of all genes only through experimental methods is challenging. A computational approach with reliable interpretability is needed to infer the function of genes, particularly for non-coding RNAs. We have analyzed genomic features that are present across both coding and non-coding genes like transcription factor (TF) and cofactor ChIP-seq (823), histone modifications ChIP-seq (n = 621), cap analysis gene expression (CAGE) tags (n = 255), and DNase hypersensitivity profiles (n = 255) to predict ontology-based functions of genes. Our approach for gene function prediction was reliable (>90% balanced accuracy) for 486 gene-sets. PubMed abstract mining and CRISPR screens supported the inferred association of genes with biological functions, for which our method had high accuracy. Further analysis revealed that TF-binding patterns at promoters have high predictive strength for multiple functions. TF-binding patterns at the promoter add an unexplored dimension of explainable regulatory aspects of genes and their functions. Therefore, we performed a comprehensive analysis for the functional-specificity of TF-binding patterns at promoters and used them for clustering functions to reveal many latent groups of gene-sets involved in common major cellular processes. We also showed how our approach could be used to infer the functions of non-coding genes using the CRISPR screens of coding genes, which were validated using a long non-coding RNA CRISPR screen. Thus our results demonstrated the generality of our approach by using gene-sets from CRISPR screens. Overall, our approach opens an avenue for predicting the involvement of non-coding genes in various functions.

FIRRM/C1orf112 is synthetic lethal with PICH and mediates RAD51 dynamics

Joint DNA molecules are natural byproducts of DNA replication and repair. Persistent joint molecules give rise to ultrafine DNA bridges (UFBs) in mitosis, compromising sister chromatid separation. The DNA translocase PICH (ERCC6L) has a central role in UFB resolution. A genome-wide loss-of-function screen is performed to identify the genetic context of PICH dependency. In addition to genes involved in DNA condensation, centromere stability, and DNA-damage repair, we identify FIGNL1-interacting regulator of recombination and mitosis (FIRRM), formerly known as C1orf112. We find that FIRRM interacts with and stabilizes the AAA⁺ ATPase FIGNL1. Inactivation of either FIRRM or FIGNL1 results in UFB formation, prolonged accumulation of RAD51 at nuclear foci, and impaired replication fork dynamics and consequently impairs genome maintenance. Combined, our data suggest that inactivation of FIRRM and FIGNL1 dysregulates RAD51 dynamics at replication forks, resulting in persistent DNA lesions and a dependency on PICH to preserve cell viability.

A global genetic interaction network by single-cell imaging and machine learning

Cellular and organismal phenotypes are controlled by complex gene regulatory networks. However, reference maps of gene function are still scarce across different organisms. Here, we generated synthetic genetic interaction and cell morphology profiles of more than 6,800 genes in cultured Drosophila cells. The resulting map of genetic interactions was used for machine learning-based gene function discovery, assigning functions to genes in 47 modules. Furthermore, we devised Cytoclass as a method to dissect genetic interactions for discrete cell states at the single-cell resolution. This approach identified an interaction of Cdk2 and the Cop9 signalosome complex, triggering senescence-associated secretory phenotypes and immunogenic conversion in hemocytic cells. Together, our data constitute a genome-scale resource of functional gene profiles to uncover the mechanisms underlying genetic interactions and their plasticity at the single-cell level.

Integrating unsupervised language model with triplet neural networks for protein gene ontology prediction

Accurate identification of protein function is critical to elucidate life mechanisms and design new drugs. We proposed a novel deep-learning method, ATGO, to predict Gene Ontology (GO) attributes of proteins through a triplet neural-network architecture embedded with pre-trained language models from protein sequences. The method was systematically tested on 1068 non-redundant benchmarking proteins and 3328 targets from the third Critical Assessment of Protein Function Annotation (CAFA) challenge. Experimental results showed that ATGO achieved a significant increase of the GO prediction accuracy compared to the state-of-the-art approaches in all aspects of molecular function, biological process, and cellular component. Detailed data analyses showed that the major advantage of ATGO lies in the utilization of pre-trained transformer language models which can extract discriminative functional pattern from the feature embeddings. Meanwhile, the proposed triplet network helps enhance the association of functional similarity with feature similarity in the sequence embedding space. In addition, it was found that the combination of the network scores with the complementary homology-based inferences could further improve the accuracy of the predicted models. These results demonstrated a new avenue for high-accuracy deep-learning function prediction that is applicable to large-scale protein function annotations from sequence alone.

TripletGO: Integrating Transcript Expression Profiles with Protein Homology Inferences for Gene Function Prediction

Gene Ontology (GO) has been widely used to annotate functions of genes and gene products. Here, we proposed a new method, TripletGO, to deduce GO terms of protein-coding and non-coding genes, through the integration of four complementary pipelines built on transcript expression profile, genetic sequence alignment, protein sequence alignment, and naïve probability. TripletGO was tested on a large set of 5754 genes from 8 species (human, mouse, Arabidopsis, rat, fly, budding yeast, fission yeast, and nematoda) and 2433 proteins with available expression data from the third Critical Assessment of Protein Function Annotation challenge (CAFA3). Experimental results show that TripletGO achieves function annotation accuracy significantly beyond the current state-of-the-art approaches. Detailed analyses show that the major advantage of TripletGO lies in the coupling of a new triplet network-based profiling method with the feature space mapping technique, which can accurately recognize function patterns from transcript expression profiles. Meanwhile, the combination of multiple complementary models, especially those from transcript expression and protein-level alignments, improves the coverage and accuracy of the final GO annotation results. The standalone package and an online server of TripletGO are freely available at https://zhanggroup.org/TripletGO/.

Integrated Analysis of Expression Profile and Potential Pathogenic Mechanism of Temporal Lobe Epilepsy With Hippocampal Sclerosis

Objective

To investigate the potential pathogenic mechanism of temporal lobe epilepsy with hippocampal sclerosis (TLE+HS) by analyzing the expression profiles of microRNA/ mRNA/ lncRNA/ DNA methylation in brain tissues.

Methods

Brain tissues of six patients with TLE+HS and nine of normal temporal or parietal cortices (NTP) of patients undergoing internal decompression for traumatic brain injury (TBI) were collected. The total RNA was dephosphorylated, labeled, and hybridized to the Agilent Human miRNA Microarray, Release 19.0, 8 × 60K. The cDNA was labeled and hybridized to the Agilent LncRNA+mRNA Human Gene Expression Microarray V3.0,4 × 180K. For methylation detection, the DNA was labeled and hybridized to the Illumina 450K Infinium Methylation BeadChip. The raw data was extracted from hybridized images using Agilent Feature Extraction, and quantile normalization was performed using the Agilent GeneSpring. P-value < 0.05 and absolute fold change >2 were considered the threshold of differential expression data. Data analyses were performed using R and Bioconductor. BrainSpan database was used to screen for signatures that were not differentially expressed in normal human hippocampus and cortex (data from BrainSpan), but differentially expressed in TLE+HS’ hippocampus and NTP’ cortex (data from our cohort). The strategy “Guilt by association” was used to predict the prospective roles of each important hub mRNA, miRNA, or lncRNA.

Results

A significantly negative correlation (r < −0.5) was found between 116 pairs of microRNA/mRNA, differentially expressed in six patients with TLE+HS and nine of NTP. We examined this regulation network’s intersection with target gene prediction results and built a lncRNA-microRNA-Gene regulatory network with structural, and functional significance. Meanwhile, we found that the disorder of FGFR3, hsa-miR-486-5p, and lnc-KCNH5-1 plays a key vital role in developing TLE+HS.

Integrative analysis of expression profile indicates the ECM receptor and LTP dysfunction in the glioma-related epilepsy

Background

Seizures are a common symptom in glioma patients, and they can cause brain dysfunction. However, the mechanism by which glioma-related epilepsy (GRE) causes alterations in brain networks remains elusive.

Objective

To investigate the potential pathogenic mechanism of GRE by analyzing the dynamic expression profiles of microRNA/ mRNA/ lncRNA in brain tissues of glioma patients.

Methods

Brain tissues of 16 patients with GRE and 9 patients with glioma without epilepsy (GNE) were collected. The total RNA was dephosphorylated, labeled, and hybridized to the Agilent Human miRNA Microarray, Release 19.0, 8 × 60 K. The cDNA was labeled and hybridized to the Agilent LncRNA + mRNA Human Gene Expression Microarray V3.0, 4 × 180 K. The raw data was extracted from hybridized images using Agilent Feature Extraction, and quantile normalization was performed using the Agilent GeneSpring. P-value < 0.05 and absolute fold change > 2 were considered the threshold of differential expression data. Data analyses were performed using R and Bioconductor.

Results

We found that 3 differentially expressed miRNAs (miR-10a-5p, miR-10b-5p, miR-629-3p), 6 differentially expressed lncRNAs (TTN-AS1, LINC00641, SNHG14, LINC00894, SNHG1, OIP5-AS1), and 49 differentially expressed mRNAs play a vitally critical role in developing GRE. The expression of GABARAPL1, GRAMD1B, and IQSEC3 were validated more than twofold higher in the GRE group than in the GNE group in the validation cohort. Pathways including ECM receptor interaction and long-term potentiation (LTP) may contribute to the disease’s progression. Meanwhile, We built a lncRNA-microRNA-Gene regulatory network with structural and functional significance.

Conclusion

These findings can offer a fresh perspective on GRE-induced brain network changes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-022-08665-8.

Validation of Novel Molecular Imaging Targets Identified by Functional Genomic mRNA Profiling to Detect Dysplasia in Barrett's Esophagus

Barrett’s esophagus (BE) is the precursor of esophageal adenocarcinoma (EAC). Dysplastic BE (DBE) has a higher progression risk to EAC compared to non-dysplastic BE (NDBE). However, the miss rates for the endoscopic detection of DBE remain high. Fluorescence molecular endoscopy (FME) can detect DBE and mucosal EAC by highlighting the tumor-specific expression of proteins. This study aimed to identify target proteins suitable for FME. Publicly available RNA expression profiles of EAC and NDBE were corrected by functional genomic mRNA (FGmRNA) profiling. Following a class comparison between FGmRNA profiles of EAC and NDBE, predicted, significantly upregulated genes in EAC were prioritized by a literature search. Protein expression of prioritized genes was validated by immunohistochemistry (IHC) on DBE and NDBE tissues. Near-infrared fluorescent tracers targeting the proteins were developed and evaluated ex vivo on fresh human specimens. In total, 1976 overexpressed genes were identified in EAC (n = 64) compared to NDBE (n = 66) at RNA level. Prioritization and IHC validation revealed SPARC, SULF1, PKCι, and DDR1 (all p < 0.0001) as the most attractive imaging protein targets for DBE detection. Newly developed tracers SULF1-800CW and SPARC-800CW both showed higher fluorescence intensity in DBE tissue compared to paired non-dysplastic tissue. This study identified SPARC, SULF1, PKCι, and DDR1 as promising targets for FME to differentiate DBE from NDBE tissue, for which SULF1-800CW and SPARC-800CW were successfully ex vivo evaluated. Clinical studies should further validate these findings.

An mRNA expression-based signature for oncogene-induced replication-stress

Oncogene-induced replication stress characterizes many aggressive cancers. Several treatments are being developed that target replication stress, however, identification of tumors with high levels of replication stress remains challenging. We describe a gene expression signature of oncogene-induced replication stress. A panel of triple-negative breast cancer (TNBC) and non-transformed cell lines were engineered to overexpress CDC25A, CCNE1 or MYC, which resulted in slower replication kinetics. RNA sequencing analysis revealed a set of 52 commonly upregulated genes. In parallel, mRNA expression analysis of patient-derived tumor samples (TCGA, n = 10,592) also revealed differential gene expression in tumors with amplification of oncogenes that trigger replication stress (CDC25A, CCNE1, MYC, CCND1, MYB, MOS, KRAS, ERBB2, and E2F1). Upon integration, we identified a six-gene signature of oncogene-induced replication stress (NAT10, DDX27, ZNF48, C8ORF33, MOCS3, and MPP6). Immunohistochemical analysis of NAT10 in breast cancer samples (n = 330) showed strong correlation with expression of phospho-RPA (R = 0.451, p = 1.82 × 10^-20) and γH2AX (R = 0.304, p = 2.95 × 10^-9). Finally, we applied our oncogene-induced replication stress signature to patient samples from TCGA (n = 8,862) and GEO (n = 13,912) to define the levels of replication stress across 27 tumor subtypes, identifying diffuse large B cell lymphoma, ovarian cancer, TNBC and colorectal carcinoma as cancer subtypes with high levels of oncogene-induced replication stress.

Lysophosphatidic acid receptor 6 regulated by miR-27a-3p attenuates tumor proliferation in breast cancer

Purpose

Lysophosphatidic acid (LPA) is a bioactive molecule which participates in many physical and pathological processes. Although LPA receptor 6 (LPAR6), the last identified LPA receptor, has been reported to have diverse effects in multiple cancers, including breast cancer, its effects and functioning mechanisms are not fully known.

Methods

Multiple public databases were used to investigate the mRNA expression of LPAR6, its prognostic value, and potential mechanisms in breast cancer. Western blotting was performed to validate the differential expression of LPAR6 in breast cancer tissues and their adjacent tissues. Furthermore, in vitro experiments were used to explore the effects of LPAR6 on breast cancer. Additionally, TargetScan and miRWalk were used to identify potential upstream regulating miRNAs and validated the relationship between miR-27a-3p and LPAR6 via real-time polymerase chain reaction and an in vitro rescue assay.

Results

LPAR6 was significantly downregulated in breast cancer at transcriptional and translational levels. Decreased LPAR6 expression in breast cancer is significantly correlated with poor overall survival, disease-free survival, and distal metastasis-free survival, particularly for hormone receptor-positive patients, regardless of lymph node metastatic status. In vitro gain and loss-of-function assays indicated that LPAR6 attenuated breast cancer cell proliferation. The analyses of TCGA and METABRIC datasets revealed that LPAR6 may regulate the cell cycle signal pathway. Furthermore, the expression of LPAR6 could be positively regulated by miR-27a-3p. The knockdown of miR-27a-3p increased cell proliferation, and ectopic expression of LPAR6 could partly rescue this phenotype.

Conclusion

LPAR6 acts as a tumor suppressor in breast cancer and is positively regulated by miR-27a-3p.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12094-021-02704-8.

Identification and Validation of Esophageal Squamous Cell Carcinoma Targets for Fluorescence Molecular Endoscopy

Dysplasia and intramucosal esophageal squamous cell carcinoma (ESCC) frequently go unnoticed with white-light endoscopy and, therefore, progress to invasive tumors. If suitable targets are available, fluorescence molecular endoscopy might be promising to improve early detection. Microarray expression data of patient-derived normal esophagus (n = 120) and ESCC samples (n = 118) were analyzed by functional genomic mRNA (FGmRNA) profiling to predict target upregulation on protein levels. The predicted top 60 upregulated genes were prioritized based on literature and immunohistochemistry (IHC) validation to select the most promising targets for fluorescent imaging. By IHC, GLUT1 showed significantly higher expression in ESCC tissue (30 patients) compared to the normal esophagus adjacent to the tumor (27 patients) (p < 0.001). Ex vivo imaging of GLUT1 with the 2-DG 800CW tracer showed that the mean fluorescence intensity in ESCC (n = 17) and high-grade dysplasia (HGD, n = 13) is higher (p < 0.05) compared to that in low-grade dysplasia (LGD) (n = 7) and to the normal esophagus adjacent to the tumor (n = 5). The sensitivity and specificity of 2-DG 800CW to detect HGD and ESCC is 80% and 83%, respectively (ROC = 0.85). We identified and validated GLUT1 as a promising molecular imaging target and demonstrated that fluorescent imaging after topical application of 2-DG 800CW can differentiate HGD and ESCC from LGD and normal esophagus.

The Increasing Impact of Translational Research in the Molecular Diagnostics of Neuromuscular Diseases

The diagnosis of neuromuscular diseases (NMDs) has been progressively evolving from the grouping of clinical symptoms and signs towards the molecular definition. Optimal clinical, biochemical, electrophysiological, electrophysiological, and histopathological characterization is very helpful to achieve molecular diagnosis, which is essential for establishing prognosis, treatment and genetic counselling. Currently, the genetic approach includes both the gene-targeted analysis in specific clinically recognizable diseases, as well as genomic analysis based on next-generation sequencing, analyzing either the clinical exome/genome or the whole exome or genome. However, as of today, there are still many patients in whom the causative genetic variant cannot be definitely established and variants of uncertain significance are often found. In this review, we address these drawbacks by incorporating two additional biological omics approaches into the molecular diagnostic process of NMDs. First, functional genomics by introducing experimental cell and molecular biology to analyze and validate the variant for its biological effect in an in-house translational diagnostic program, and second, incorporating a multi-omics approach including RNA-seq, metabolomics, and proteomics in the molecular diagnosis of neuromuscular disease. Both translational diagnostics programs and omics are being implemented as part of the diagnostic process in academic centers and referral hospitals and, therefore, an increase in the proportion of neuromuscular patients with a molecular diagnosis is expected. This improvement in the process and diagnostic performance of patients will allow solving aspects of their health problems in a precise way and will allow them and their families to take a step forward in their lives.

Improving gene function predictions using independent transcriptional components

The interpretation of high throughput sequencing data is limited by our incomplete functional understanding of coding and non-coding transcripts. Reliably predicting the function of such transcripts can overcome this limitation. Here we report the use of a consensus independent component analysis and guilt-by-association approach to predict over 23,000 functional groups comprised of over 55,000 coding and non-coding transcripts using publicly available transcriptomic profiles. We show that, compared to using Principal Component Analysis, Independent Component Analysis-derived transcriptional components enable more confident functionality predictions, improve predictions when new members are added to the gene sets, and are less affected by gene multi-functionality. Predictions generated using human or mouse transcriptomic data are made available for exploration in a publicly available web portal.