how Lordec maps the long reads to DeBruijn Graph

Question

[midox]

hello,

I would like how LorDEC maps long reads to debruijn graph. After construction of deBruijn graph of short reads how map the set of erroneous long reads to DeBuijn graph. it's not clear in the paper.

can you help me please?

Thank you for your response.

Answer 1

[Medhat]

As far as I understand, after forming DBG graphs from short reads (SR),the use and store data in GATB library which allows to traverse any path in the graph and to get the sequence of any node. also PacBio long reads (LR) are divided to k-mer which will be compared (traversed) to the k-mer in the DBG of SR.

They used solid k-mer as anchor to correct PacBio reads

We found that if we additionally require that for a k-mer to be considered solid, it must also have at least one incoming and at least one outgoing arc

how they correct the reads?

Consider the k-mers of a long read starting at position 1,2,3, … : some k-mers belong to the graph and are solid, while others do not and are weak. Basically, solid k-mers are expected to be correct, while weak ones suspectedly include sequencing errors and require a correction. Solid k-mers are entry points in the DBG, and LoRDEC corrects a region made of weak k-mers by finding the best path in the DBG between the solid k-mers bordering this region. Sometimes, an LR has no solid k-mer, in which case, LoRDEC marks it as such in the output and skips it.

The idea here is traversing not mapping (as I understand ) for more about traversing graph traverse

and here explanation about how it works

http://ivory.idyll.org/blog/2015-wok-error-correction.html

and here is a paper talking about the mapping of sequence to DBG

Read mapping on de Bruijn graphs

Answer 2

[midox]

thank you for your response. And LORMA? have you an explanation about LORMA? Thanks

[Medhat]

It is divided into two phases :

1. first using lordec with different k-mer to self coorect the long
2. reads second further correction with the multiple alignment technique by LoRMA

[midox]

have more details in this phases? because it's not clear in the paper. Thanks

[Medhat]

unfortunately I do not have enough info