Pathformer: a biological pathway informed transformer for disease diagnosis and prognosis using multi-omics data.

MOTIVATION: Multi-omics data provide a comprehensive view of gene regulation at multiple levels, which is helpful in achieving accurate diagnosis of complex diseases like cancer. However, conventional integration methods rarely utilize prior biological knowledge and lack interpretability. RESULTS: To integrate various multi-omics data of tissue and liquid biopsies for disease diagnosis and prognosis, we developed a biological pathway informed Transformer, Pathformer. It embeds multi-omics input with a compacted multi-modal vector and a pathway-based sparse neural network. Pathformer also leverages criss-cross attention mechanism to capture the crosstalk between different pathways and modalities. We first benchmarked Pathformer with 18 comparable methods on multiple cancer datasets, where Pathformer outperformed all the other methods, with an average improvement of 6.3%-14.7% in F1 score for cancer survival prediction, 5.1%-12% for cancer stage prediction, and 8.1%-13.6% for cancer drug response prediction. Subsequently, for cancer prognosis prediction based on tissue multi-omics data, we used a case study to demonstrate the biological interpretability of Pathformer by identifying key pathways and their biological crosstalk. Then, for cancer early diagnosis based on liquid biopsy data, we used plasma and platelet datasets to demonstrate Pathformer's potential of clinical applications in cancer screening. Moreover, we revealed deregulation of interesting pathways (e.g. scavenger receptor pathway) and their crosstalk in cancer patients' blood, providing potential candidate targets for cancer microenvironment study. AVAILABILITY AND IMPLEMENTATION: Pathformer is implemented and freely available at https://github.com/lulab/Pathformer.

cfOmics: a cell-free multi-Omics database for diseases.

Liquid biopsy has emerged as a promising non-invasive approach for detecting, monitoring diseases, and predicting their recurrence. However, the effective utilization of liquid biopsy data to identify reliable biomarkers for various cancers and other diseases requires further exploration. Here, we present cfOmics, a web-accessible database (https://cfomics.ncRNAlab.org/) that integrates comprehensive multi-omics liquid biopsy data, including cfDNA, cfRNA based on next-generation sequencing, and proteome, metabolome based on mass-spectrometry data. As the first multi-omics database in the field, cfOmics encompasses a total of 17 distinct data types and 13 specimen variations across 69 disease conditions, with a collection of 11345 samples. Moreover, cfOmics includes reported potential biomarkers for reference. To facilitate effective analysis and visualization of multi-omics data, cfOmics offers powerful functionalities to its users. These functionalities include browsing, profile visualization, the Integrative Genomic Viewer, and correlation analysis, all centered around genes, microbes, or end-motifs. The primary objective of cfOmics is to assist researchers in the field of liquid biopsy by providing comprehensive multi-omics data. This enables them to explore cell-free data and extract profound insights that can significantly impact disease diagnosis, treatment monitoring, and management.

Cell-free multi-omics analysis reveals potential biomarkers in gastrointestinal cancer patients' blood.

During cancer progression, tumorigenic and immune signals are spread through circulating molecules, such as cell-free DNA (cfDNA) and cell-free RNA (cfRNA) in the blood. So far, they have not been comprehensively investigated in gastrointestinal cancers. Here, we profile 4 categories of cell-free omics data from patients with colorectal cancer and patients with stomach adenocarcinoma and then assay 15 types of genomic, epigenomic, and transcriptomic variations. We find that multi-omics data are more appropriate for detection of cancer genes compared with single-omics data. In particular, cfRNAs are more sensitive and informative than cfDNAs in terms of detection rate, enriched functional pathways, etc. Moreover, we identify several peripheral immune signatures that are suppressed in patients with cancer. Specifically, we establish a γδ-T cell score and a cancer-associated-fibroblast (CAF) score, providing insights into clinical statuses like cancer stage and survival. Overall, we reveal a cell-free multi-molecular landscape that is useful for blood monitoring in personalized cancer treatment.

A comprehensive evaluation of full-spectrum cell-free RNAs highlights cell-free RNA fragments for early-stage hepatocellular carcinoma detection.

BACKGROUND: Various studies have reported cell-free RNAs (cfRNAs) as noninvasive biomarkers for detecting hepatocellular carcinoma (HCC). However, they have not been independently validated, and some results are contradictory. We provided a comprehensive evaluation of various types of cfRNA biomarkers and a full mining of the biomarker potential of new features of cfRNA. METHODS: We first systematically reviewed reported cfRNA biomarkers and calculated dysregulated post-transcriptional events and cfRNA fragments. In 3 independent multicentre cohorts, we further selected 6 cfRNAs using RT-qPCR, built a panel called HCCMDP with AFP using machine learning, and internally and externally validated HCCMDP's performance. FINDINGS: We identified 23 cfRNA biomarker candidates from a systematic review and analysis of 5 cfRNA-seq datasets. Notably, we defined the cfRNA domain to describe cfRNA fragments systematically. In the verification cohort (n = 183), cfRNA fragments were more likely to be verified, while circRNA and chimeric RNA candidates were neither abundant nor stable as qPCR-based biomarkers. In the algorithm development cohort (n = 287), we build and test the panel HCCMDP with 6 cfRNA markers and AFP. In the independent validation cohort (n = 171), HCCMDP can distinguish HCC patients from control groups (all: AUC = 0.925; CHB: AUC = 0.909; LC: AUC = 0.916), and performs well in distinguishing early-stage HCC patients (all: AUC = 0.936; CHB: AUC = 0.917; LC: AUC = 0.928). INTERPRETATION: This study comprehensively evaluated full-spectrum cfRNA biomarker types for HCC detection, highlighted the cfRNA fragment as a promising biomarker type in HCC detection, and provided a panel HCCMDP. FUNDING: National Natural Science Foundation of China, and The National Key Basic Research Program (973 program).

Cell-free RNA for the liquid biopsy of gastrointestinal cancer.

Gastrointestinal (GI) cancer includes many cancer types, such as esophageal, liver, gastric, pancreatic, and colorectal cancer. As the cornerstone of personalized medicine for GI cancer, liquid biopsy based on noninvasive biomarkers provides promising opportunities for early diagnosis and dynamic treatment management. Recently, a growing number of studies have demonstrated the potential of cell-free RNA (cfRNA) as a new type of noninvasive biomarker in body fluids, such as blood, saliva, and urine. Meanwhile, transcriptomes based on high-throughput RNA detection technologies keep discovering new cfRNA biomarkers. In this review, we introduce the origins and applications of cfRNA, describe its detection and qualification methods in liquid biopsy, and summarize a comprehensive list of cfRNA biomarkers in different GI cancer types. Moreover, we also discuss perspective studies of cfRNA to overcome its current limitations in clinical applications. This article is categorized under: RNA in Disease and Development > RNA in Disease.

Long non-coding RNA DARS-AS1 promotes tumor progression by directly suppressing PACT-mediated cellular stress.

Cancer cells evolve various mechanisms to overcome cellular stresses and maintain progression. Protein kinase R (PKR) and its protein activator (PACT) are the initial responders in monitoring diverse stress signals and lead to inhibition of cell proliferation and cell apoptosis in consequence. However, the regulation of PACT-PKR pathway in cancer cells remains largely unknown. Herein, we identify that the long non-coding RNA (lncRNA) aspartyl-tRNA synthetase antisense RNA 1 (DARS-AS1) is directly involved in the inhibition of the PACT-PKR pathway and promotes the proliferation of cancer cells. Using large-scale CRISPRi functional screening of 971 cancer-associated lncRNAs, we find that DARS-AS1 is associated with significantly enhanced proliferation of cancer cells. Accordingly, knocking down DARS-AS1 inhibits cell proliferation of multiple cancer cell lines and promotes cancer cell apoptosis in vitro and significantly reduces tumor growth in vivo. Mechanistically, DARS-AS1 directly binds to the activator domain of PACT and prevents PACT-PKR interaction, thereby decreasing PKR activation, eIF2α phosphorylation and inhibiting apoptotic cell death. Clinically, DARS-AS1 is broadly expressed across multiple cancers and the increased expression of this lncRNA indicates poor prognosis. This study elucidates the lncRNA DARS-AS1 directed cancer-specific modulation of the PACT-PKR pathway and provides another target for cancer prognosis and therapeutic treatment.

Cancer type classification using plasma cell-free RNAs derived from human and microbes.

The utility of cell-free nucleic acids in monitoring cancer has been recognized by both scientists and clinicians. In addition to human transcripts, a fraction of cell-free nucleic acids in human plasma were proven to be derived from microbes and reported to have relevance to cancer. To obtain a better understanding of plasma cell-free RNAs (cfRNAs) in cancer patients, we profiled cfRNAs in ~300 plasma samples of 5 cancer types (colorectal cancer, stomach cancer, liver cancer, lung cancer, and esophageal cancer) and healthy donors (HDs) with RNA-seq. Microbe-derived cfRNAs were consistently detected by different computational methods when potential contaminations were carefully filtered. Clinically relevant signals were identified from human and microbial reads, and enriched Kyoto Encyclopedia of Genes and Genomes pathways of downregulated human genes and higher prevalence torque teno viruses both suggest that a fraction of cancer patients were immunosuppressed. Our data support the diagnostic value of human and microbe-derived plasma cfRNAs for cancer detection, as an area under the ROC curve of approximately 0.9 for distinguishing cancer patients from HDs was achieved. Moreover, human and microbial cfRNAs both have cancer type specificity, and combining two types of features could distinguish tumors of five different primary locations with an average recall of 60.4%. Compared to using human features alone, adding microbial features improved the average recall by approximately 8%. In summary, this work provides evidence for the clinical relevance of human and microbe-derived plasma cfRNAs and their potential utilities in cancer detection as well as the determination of tumor sites.

A Nodal enhanced micropeptide NEMEP regulates glucose uptake during mesendoderm differentiation of embryonic stem cells.

TGF-ß family proteins including Nodal are known as central regulators of early development in metazoans, yet our understanding of the scope of Nodal signaling's downstream targets and associated physiological mechanisms in specifying developmentally appropriate cell fates is far from complete. Here, we identified a highly conserved, transmembrane micropeptide-NEMEP-as a direct target of Nodal signaling in mesendoderm differentiation of mouse embryonic stem cells (mESCs), and this micropeptide is essential for mesendoderm differentiation. We showed that NEMEP interacts with the glucose transporters GLUT1/GLUT3 and promotes glucose uptake likely through these interactions. Thus, beyond expanding the scope of known Nodal signaling targets in early development and showing that this target micropeptide augments the glucose uptake during mesendoderm differentiation, our study provides a clear example for the direct functional impact of altered glucose metabolism on cell fate determination.

POSTAR3: an updated platform for exploring post-transcriptional regulation coordinated by RNA-binding proteins.

RNA-binding proteins (RBPs) play key roles in post-transcriptional regulation. Accurate identification of RBP binding sites in multiple cell lines and tissue types from diverse species is a fundamental endeavor towards understanding the regulatory mechanisms of RBPs under both physiological and pathological conditions. Our POSTAR annotation processes make use of publicly available large-scale CLIP-seq datasets and external functional genomic annotations to generate a comprehensive map of RBP binding sites and their association with other regulatory events as well as functional variants. Here, we present POSTAR3, an updated database with improvements in data collection, annotation infrastructure, and analysis that support the annotation of post-transcriptional regulation in multiple species including: we made a comprehensive update on the CLIP-seq and Ribo-seq datasets which cover more biological conditions, technologies, and species; we added RNA secondary structure profiling for RBP binding sites; we provided miRNA-mediated degradation events validated by degradome-seq; we included RBP binding sites at circRNA junction regions; we expanded the annotation of RBP binding sites, particularly using updated genomic variants and mutations associated with diseases. POSTAR3 is freely available at http://postar.ncrnalab.org.

Methylome profiling identifies TCHH methylation in CfDNA as a noninvasive marker of liver metastasis in colorectal cancer.

Methylation of circulating free DNA (CfDNA) has emerged as an efficient marker of tumor screening and prognostics. However, no efficient methylation marker has been developed for monitoring liver metastasis (LM) in colorectal cancer (CRC). Utilizing methylome profiling and bisulfite sequencing polymerase chain reaction of paired primary and LM sites, significantly increased methylation of TCHH was identified in the process of LM in CRC in the present study. Methylight analysis of TCHH methylation in CfDNA displayed a promisingly discriminative power between CRC with and without LM. Besides, significant coefficient of TCHH methylation and LM tumor volume was also validated. Together, these results indicated the potential of TCHH methylation in CfDNA as a monitoring marker of LM in CRC.

Integrative genomic analysis of early neurogenesis reveals a temporal genetic program for differentiation and specification of preplate and Cajal-Retzius neurons.

Neurogenesis in the developing neocortex begins with the generation of the preplate, which consists of early-born neurons including Cajal-Retzius (CR) cells and subplate neurons. Here, utilizing the Ebf2-EGFP transgenic mouse in which EGFP initially labels the preplate neurons then persists in CR cells, we reveal the dynamic transcriptome profiles of early neurogenesis and CR cell differentiation. Genome-wide RNA-seq and ChIP-seq analyses at multiple early neurogenic stages have revealed the temporal gene expression dynamics of early neurogenesis and distinct histone modification patterns in early differentiating neurons. We have identified a new set of coding genes and lncRNAs involved in early neuronal differentiation and validated with functional assays in vitro and in vivo. In addition, at E15.5 when Ebf2-EGFP+ cells are mostly CR neurons, single-cell sequencing analysis of purified Ebf2-EGFP+ cells uncovers molecular heterogeneities in CR neurons, but without apparent clustering of cells with distinct regional origins. Along a pseudotemporal trajectory these cells are classified into three different developing states, revealing genetic cascades from early generic neuronal differentiation to late fate specification during the establishment of CR neuron identity and function. Our findings shed light on the molecular mechanisms governing the early differentiation steps during cortical development, especially CR neuron differentiation.

Integrative analysis of long extracellular RNAs reveals a detection panel of noncoding RNAs for liver cancer.

Rationale: Long extracellular RNAs (exRNAs) in plasma can be profiled by new sequencing technologies, even with low abundance. However, cancer-related exRNAs and their variations remain understudied. Methods: We investigated different variations (i.e. differential expression, alternative splicing, alternative polyadenylation, and differential editing) in diverse long exRNA species (e.g. long noncoding RNAs and circular RNAs) using 79 plasma exosomal RNA-seq (exoRNA-seq) datasets of multiple cancer types. We then integrated 53 exoRNA-seq datasets and 65 self-profiled cell-free RNA-seq (cfRNA-seq) datasets to identify recurrent variations in liver cancer patients. We further combined TCGA tissue RNA-seq datasets and validated biomarker candidates by RT-qPCR in an individual cohort of more than 100 plasma samples. Finally, we used machine learning models to identify a signature of 3 noncoding RNAs for the detection of liver cancer. Results: We found that different types of RNA variations identified from exoRNA-seq data were enriched in pathways related to tumorigenesis and metastasis, immune, and metabolism, suggesting that cancer signals can be detected from long exRNAs. Subsequently, we identified more than 100 recurrent variations in plasma from liver cancer patients by integrating exoRNA-seq and cfRNA-seq datasets. From these datasets, 5 significantly up-regulated long exRNAs were confirmed by TCGA data and validated by RT-qPCR in an independent cohort. When using machine learning models to combine two of these validated circular and structured RNAs (SNORD3B-1, circ-0080695) with a miRNA (miR-122) as a panel to classify liver cancer patients from healthy donors, the average AUROC of the cross-validation was 89.4%. The selected 3-RNA panel successfully detected 79.2% AFP-negative samples and 77.1% early-stage liver cancer samples in the testing and validation sets. Conclusions: Our study revealed that different types of RNA variations related to cancer can be detected in plasma and identified a 3-RNA detection panel for liver cancer, especially for AFP-negative and early-stage patients.

Arabidopsis exoribonuclease USB1 interacts with the PPR-domain protein SOAR1 to negatively regulate abscisic acid signaling.

Signaling by the phytohormone abscisic acid (ABA) involves pre-mRNA splicing, a key process of post-transcriptional regulation of gene expression. However, the regulatory mechanism of alternative pre-mRNA splicing in ABA signaling remains largely unknown. We previously identified a pentatricopeptide repeat protein SOAR1 (suppressor of the ABAR-overexpressor 1) as a crucial player downstream of ABAR (putative ABA receptor) in ABA signaling. In this study, we identified a SOAR1 interaction partner USB1, which is an exoribonuclease catalyzing U6 production for spliceosome assembly. We reveal that together USB1 and SOAR1 negatively regulate ABA signaling in early seedling development. USB1 and SOAR1 are both required for the splicing of transcripts of numerous genes, including those involved in ABA signaling pathways, suggesting that USB1 and SOAR1 collaborate to regulate ABA signaling by affecting spliceosome assembly. These findings provide important new insights into the mechanistic control of alternative pre-mRNA splicing in the regulation of ABA-mediated plant responses to environmental cues.

Selective translation by alternative bacterial ribosomes.

Alternative ribosome subunit proteins are prevalent in the genomes of diverse bacterial species, but their functional significance is controversial. Attempts to study microbial ribosomal heterogeneity have mostly relied on comparing wild-type strains with mutants in which subunits have been deleted, but this approach does not allow direct comparison of alternate ribosome isoforms isolated from identical cellular contexts. Here, by simultaneously purifying canonical and alternative RpsR ribosomes from Mycobacterium smegmatis, we show that alternative ribosomes have distinct translational features compared with their canonical counterparts. Both alternative and canonical ribosomes actively take part in protein synthesis, although they translate a subset of genes with differential efficiency as measured by ribosome profiling. We also show that alternative ribosomes have a relative defect in initiation complex formation. Furthermore, a strain of M. smegmatis in which the alternative ribosome protein operon is deleted grows poorly in iron-depleted medium, uncovering a role for alternative ribosomes in iron homeostasis. Our work confirms the distinct and nonredundant contribution of alternative bacterial ribosomes for adaptation to hostile environments.

Tissue-specific transcription reprogramming promotes liver metastasis of colorectal cancer.

Metastasis, the development of secondary malignant growths at a distance from a primary tumor, is the cause of death for 90% of cancer patients, but little is known about how metastatic cancer cells adapt to and colonize new tissue environments. Here, using clinical samples, patient-derived xenograft (PDX) samples, PDX cells, and primary/metastatic cell lines, we discovered that liver metastatic colorectal cancer (CRC) cells lose their colon-specific gene transcription program yet gain a liver-specific gene transcription program. We showed that this transcription reprogramming is driven by a reshaped epigenetic landscape of both typical enhancers and super-enhancers. Further, we identified that the liver-specific transcription factors FOXA2 and HNF1A can bind to the gained enhancers and activate the liver-specific gene transcription, thereby driving CRC liver metastasis. Importantly, similar transcription reprogramming can be observed in multiple cancer types. Our data suggest that reprogrammed tissue-specific transcription promotes metastasis and should be targeted therapeutically.

A comprehensive evaluation of connectivity methods for L1000 data.

The methodologies for evaluating similarities between gene expression profiles of different perturbagens are the key to understanding mechanisms of actions (MoAs) of unknown compounds and finding new indications for existing drugs. L1000-based next-generation Connectivity Map (CMap) data is more than a thousand-fold scale-up of the CMap pilot dataset. Although several systematic evaluations have been performed individually to assess the accuracy of the methodologies for the CMap pilot study, the performance of these methodologies needs to be re-evaluated for the L1000 data. Here, using the drug-drug similarities from the Drug Repurposing Hub database as a benchmark standard, we evaluated six popular published methods for the prediction performance of drug-drug relationships based on the partial area under the receiver operating characteristic (ROC) curve at false positive rates of 0.001, 0.005 and 0.01 (AUC0.001, AUC0.005 and AUC0.01). The similarity evaluating algorithm called ZhangScore was generally superior to other methods and exhibited the highest accuracy at the gene signature sizes ranging from 10 to 200. Further, we tested these methods with an experimentally derived gene signature related to estrogen in breast cancer cells, and the results confirmed that ZhangScore was more accurate than other methods. Moreover, based on scoring results of ZhangScore for the gene signature of TOP2A knockdown, in addition to well-known TOP2A inhibitors, we identified a number of potential inhibitors and at least two of them were the subject of previous investigation. Our studies provide potential guidelines for researchers to choose the suitable connectivity method. The six connectivity methods used in this report have been implemented in R package (https://github.com/Jasonlinchina/RCSM).

Pervasive Chromatin-RNA Binding Protein Interactions Enable RNA-Based Regulation of Transcription.

Increasing evidence suggests that transcriptional control and chromatin activities at large involve regulatory RNAs, which likely enlist specific RNA-binding proteins (RBPs). Although multiple RBPs have been implicated in transcription control, it has remained unclear how extensively RBPs directly act on chromatin. We embarked on a large-scale RBP ChIP-seq analysis, revealing widespread RBP presence in active chromatin regions in the human genome. Like transcription factors (TFs), RBPs also show strong preference for hotspots in the genome, particularly gene promoters, where their association is frequently linked to transcriptional output. Unsupervised clustering reveals extensive co-association between TFs and RBPs, as exemplified by YY1, a known RNA-dependent TF, and RBM25, an RBP involved in splicing regulation. Remarkably, RBM25 depletion attenuates all YY1-dependent activities, including chromatin binding, DNA looping, and transcription. We propose that various RBPs may enhance network interaction through harnessing regulatory RNAs to control transcription.

Microarray is an efficient tool for circRNA profiling.

Circular RNAs (circRNAs) are emerging as a new class of endogenous and regulatory noncoding RNAs in latest years. With the widespread application of RNA sequencing (RNA-seq) technology and bioinformatics prediction, large numbers of circRNAs have been identified. However, at present, we lack a comprehensive characterization of all these circRNAs in interested samples. In this study, we integrated 87 935 circRNAs sequences that cover most of circRNAs identified till now represented in circBase to design microarray probes targeting back-splice site of each circRNA to profile expression of those circRNAs. By comparing the circRNA detection efficiency of RNA-seq with this circRNA microarray, we revealed that microarray is more efficient than RNA-seq for circRNA profiling. Then, we found ∼80 000 circRNAs were expressed in cervical tumors and matched normal tissues, and ∼25 000 of them were differently expressed. Notably, many of these circRNAs detected by this microarray can be validated by quantitative reverse transcription polymerase chain reaction (RT-qPCR) or RNA-seq. Strikingly, as many as ∼18 000 circRNAs could be robustly detected in cell-free plasma samples, and the expression of ∼2700 of them differed after surgery for tumor removal. Our findings provided a comprehensive and genome-wide characterization of circRNAs in paired normal tissues and tumors and plasma samples from multiple individuals. In addition, we also provide a rich resource with 41 microarray data sets and 10 RNA-seq data sets and strong evidences for circRNA expression in cervical cancer. In conclusion, circRNAs could be efficiently profiled by circRNA microarray to target their reported back-splice sites in interested samples.

POSTAR2: deciphering the post-transcriptional regulatory logics.

Post-transcriptional regulation of RNAs is critical to the diverse range of cellular processes. The volume of functional genomic data focusing on post-transcriptional regulation logics continues to grow in recent years. In the current database version, POSTAR2 (http://lulab.life.tsinghua.edu.cn/postar), we included the following new features and data: updated ∼500 CLIP-seq datasets (∼1200 CLIP-seq datasets in total) from six species, including human, mouse, fly, worm, Arabidopsis and yeast; added a new module 'Translatome', which is derived from Ribo-seq datasets and contains ∼36 million open reading frames (ORFs) in the genomes from the six species; updated and unified post-transcriptional regulation and variation data. Finally, we improved web interfaces for searching and visualizing protein-RNA interactions with multi-layer information. Meanwhile, we also merged our CLIPdb database into POSTAR2. POSTAR2 will help researchers investigate the post-transcriptional regulatory logics coordinated by RNA-binding proteins and translational landscape of cellular RNAs.