Tiglon

Tiglon performs genome-guided transcriptome assembly from RNA-seq data by integrating Hisat2 and Star alignments to reconstruct full-length transcript isoforms.

Key Features:

• Genome-guided assembly: Performs transcriptome assembly using a reference genome to guide reconstruction of transcript isoforms from RNA-seq data.
• Multi-aligner integration: Leverages mappings generated by Hisat2 and Star and can use either aligner individually or both in combination.
• Labeled splice graphs: Constructs labeled splice graphs as a structural representation of potential transcript isoforms derived from alignments.
• Label-based dynamic path-searching: Applies a label-based dynamic path-searching strategy to traverse labeled splice graphs and identify optimal transcript paths.
• Merge of mapping insights: Merges diverse mapping insights from multiple aligners to create a comprehensive framework for assembly.
• Full-length and low-expression recovery: Targets reconstruction of full-length transcripts and improves recovery of lowly expressed transcripts.
• Benchmarking and evaluation: Evaluated on simulated datasets and 12 real-world datasets with reported performance comparisons to StringTie2 and Scallop.

Scientific Applications:

• Full-length transcript reconstruction: Reconstructs complete transcript isoforms from RNA-seq alignments using genome guidance.
• Isoform and splice analysis: Identifies alternative splicing isoforms via traversal of labeled splice graphs.
• Detection of lowly expressed transcripts: Enhances recovery of transcripts expressed at low levels that are challenging for other assemblers.
• Comparative assembler evaluation: Facilitates benchmarking of assembler performance against tools such as StringTie2 and Scallop using simulated and real datasets.

Methodology:

Constructs labeled splice graphs from Hisat2 and Star mappings and applies a label-based dynamic path-searching strategy to identify optimal paths corresponding to full-length transcripts, supporting Hisat2, Star, or combined alignments.