Suggestion for hybrid error correction

[tuannguyen8390]

Hi there,

Found your wrapper over Twitter, great incentive :). I have a suggestion for your pipeline - it would be of interest to consider hybrid correction (aka. combine short + longread). With my current pipeline I was using fmlrc, combine with ONT works pretty well

Cheers,

Tuan

[fmalmeida]

Hi @tuannguyen8390,

Thank you for the feedback 😄

The suggestion you've given is totally something we also have in mind. We are just "postponing" it a little bit more because it would require to change how the pipeline reads the inputs.

For now, it reads the inputs of different techs in a total independent way, meaning they have no relation to each other and are analysed randomly since we load them with the nextflow's Channel.from() function. Which, for instance, would make it problematic to keep sure that the correct short reads are being used to correct it's related long reads when users load more than one sample at a time.

For this to work we would have to change how inputs are given to something more similar to a samplesheet in CSV/TSV so we are sure about which read to use to correct the long reads. We'd need something like that:

sample,ont,fastq_1,fastq_2
sample1,ont.fq.gz,reads_1.fq.gz,reads_2.fq.gz

Although these changes sound simple, they are a little bit trickier to implement. So yes, this is definitely on the list but it may take some time 😅. Also, we are very open to contributions, if you few comfortable with we'd very much welcome some help 👨🏼‍💻