How ablab/isoquant works
Main Components
a) Input Processing Layer:
• Multiple input sources (FASTQ/FASTA files, BAM files)
• Reference genome (mandatory)
• Optional gene annotation (GFF/GTF)
• Read grouping information
b) Core Processing Pipeline:
• Read Mapping Component (using minimap2)
• Alignment Processing
• Transcript Model Construction
• Isoform Assignment
• Quantification Engine
c) Output Generation Layer:
• Transcript models
• Gene/isoform quantification
• Expression tables
• Statistics and logs
Component Relationships
[Input Layer]
• ↓
[Preprocessing]
• ↓
[Core Processing]
• ↓
[Output Layer]
Specific Implementation Details
a) Input Processing (Blue Components):
• Parallel input paths for FASTQ/FASTA and BAM files
• Reference data input (genome, annotation)
• To the read mapper component
b) Core Processing (Green Components):
• Alignment_processor module
• Gene_model and isoform_assignment modules
• Multimap_resolver component
• Long_read_counter module
c) Analysis Modules (Yellow Components):
• Exon_corrector and illumina_exon_corrector
• Polya_finder and verification
• Graph_based_model_construction
d) Output Generation (Purple Components):
• Transcript_printer module
• Statistics generation
• Visualization components
Technical Details
• Python 3.8 requirement
• External dependencies (gffutils, pysam, pybedtools)
• File format handling capabilities
• Database interaction points
Component Grouping
Related components using containers:
• Input Processing Container
• Alignment Processing Container
• Model Construction Container
• Quantification Container
• Output Generation Container
Additional Elements
• Version information (3.6)
• Supported data types (PacBio CCS, ONT dRNA/cDNA)
• Configuration points
• Error handling paths
Remember to highlight the modular nature of the system, showing how components like the exon corrector can be optionally included in the pipeline.
This design should reflect IsoQuant's primary purpose of analyzing long RNA reads and producing accurate transcript models and quantification results.