The breakdown is roughly:
- Index processing:
- checking fastq headers, extracting indices
- testing indices for matches (or near-matches)
- calculating edit distances
- keeping counts of observed indices
- Sample sheet processing:
- parsing, storing sample sheet
- testing indices to see which destination they match
- Fastq:
- reading, storing fastq (possibly compressed source)
- extracting indices, according to pattern
- writing fastq to appropriate output streams (possibly compressed).
- Command-line processing.
- Utilities:
- logging
- determining the line-ending style of files
The code is separated into a common directory with, you guessed it, shared code, and specific directories for individual programs. So far there are only two programs:
- demuxFQ -- demultiplex fastq files, and
- indices -- examine and report on UMI and index sequences.
Common classes include:
| Class | Description | Status |
|---|---|---|
| Destination | File to write demultiplexed results | pre |
| Fastq | A fastq sequence | pre |
| FastqReader | A reader and parser for fastq-formatted files | pre |
| Index | An index (fast match with threshold to other index) | in progress |
| IndexSingle | A single index | in progress |
| IndexDual | A dual index | in progress |
| IndexPattern | A pattern to parse indices with | in progress |
| IndexSet | A set of indices | in progress |
| LineType | Utility to test line endings in a file | tested |
| Log | Utility to log messages (currently in util.cpp) | pre |
| SampleSheet | A sample sheet for demultiplexing | in progress |
IndexSet -- set of known indices (from fastq)
- factory class for making new indices
- implement a fast "seen" test
- subclasses for set of sample sheet indices, FQ file indices
- subclasses provide parsing for two cases
- ensure duplicate index -> same object
- If the sample sheet lists hundreds of indices, are we going to run into trouble with too many open file handles?