Hello,
We are working on a 454 de novo assembly of transcriptome data using CLC's de Bruijn graphs method.
We are encountering two major problems when we look at the assembly and were hoping that someone might know the answers.
1. There are a large number of N's being produced in general, however, there are some contigs that have two regions of high coverage (5-20 reads), that are connected in the consensus sequence by a long string of N's (>10), and in the assembly of the reads, there are gaps of 3-5bp where there is actually no read. Essentially the two regions of high coverage do not appear connected and are in fact two separate contigs.
2. At the either end of the contigs, there is data missing in the consensus sequence, where there is actually a collection of well-aligned reads. Why is CLC rejecting these reads as part of the contig?
For question 1, I am not sure of the answer but I am guessing it is an artifact of the De Bruijn graph method that I am not fully comprehending. Any help would be greatly appreciated.
For question 2, I believe the answer may be to do with CLC mapping the original reads back to the contigs and if the mapping is poor, then they are removed from the consensus sequence. (this was discussed in a different post).
Any help will be greatly appreciated, apologies for the length of the post.
We are working on a 454 de novo assembly of transcriptome data using CLC's de Bruijn graphs method.
We are encountering two major problems when we look at the assembly and were hoping that someone might know the answers.
1. There are a large number of N's being produced in general, however, there are some contigs that have two regions of high coverage (5-20 reads), that are connected in the consensus sequence by a long string of N's (>10), and in the assembly of the reads, there are gaps of 3-5bp where there is actually no read. Essentially the two regions of high coverage do not appear connected and are in fact two separate contigs.
2. At the either end of the contigs, there is data missing in the consensus sequence, where there is actually a collection of well-aligned reads. Why is CLC rejecting these reads as part of the contig?
For question 1, I am not sure of the answer but I am guessing it is an artifact of the De Bruijn graph method that I am not fully comprehending. Any help would be greatly appreciated.
For question 2, I believe the answer may be to do with CLC mapping the original reads back to the contigs and if the mapping is poor, then they are removed from the consensus sequence. (this was discussed in a different post).
Any help will be greatly appreciated, apologies for the length of the post.
Comment