groHMM

groHMM identifies transcriptional units from Global Run-On sequencing (GRO-seq) data, detecting coding and non-coding primary transcripts such as primary microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and enhancer RNAs (eRNAs) to map the transcriptional landscape.

Key Features:

• De Novo Transcription Unit Boundary Definition: groHMM employs a two-state hidden Markov model to define transcription unit boundaries from GRO-seq data without prior annotation.
• Modeling of GRO-seq Signal: The method models the sequential nature of GRO-seq data to segment genomic regions into active and inactive transcriptional states.
• Transcript Class Detection: groHMM detects both coding and non-coding primary transcripts, including primary microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and enhancer RNAs (eRNAs).
• Implementation: The software is implemented in R for integration into computational workflows.
• Benchmarking: groHMM has been compared against peak-calling methods SICER and HOMER on GRO-seq data from MCF-7 breast cancer cells.
• Broad Applicability: The approach has been applied to non-transformed human cells (cardiomyocytes, lung fibroblasts), transformed human cancer cells (LNCaP, MCF-7), and non-mammalian cells (flies, worms).

Scientific Applications:

• Annotation of Transcription Units: Identifying primary transcripts across diverse cell types to support comprehensive transcriptome annotation.
• Cell Type-Specific Enhancer Analysis: Annotating enhancer transcripts to analyze cell type-specific enhancer activity.
• Discovery of Novel Transcripts and Elements: Enabling identification of novel transcription units and previously unannotated functional genomic elements.
• Transcriptional Dynamics and Regulatory Mechanisms: Supporting studies of transcriptional dynamics and gene regulation by delineating active transcriptional regions.

Methodology:

Uses a two-state hidden Markov model (HMM) to segment GRO-seq signal into active and inactive states, with the implementation provided in R.