Pile up reads without reference?

Question

[pbigbig]

Hi everyone,

I have multiple groups of reads (count < 1000) which derived from a number of novel junctions, with each reads contain minimum ~10bp overhang on the other side (i.e., exon). Because the number of random links between exon-exon can be quite large (hundreds), it is difficult to construct artificial reference for every case.

I would like to ask, is the any simple way to pile up reads (fasta or fastq) and visualize single consensus sequence?. I thought of assembling them to make contigs, then map them using contigs as reference, and visualize with IGV, but there should be a simpler way to just pile them up and visualize it?

Thank you in advance for your suggestion!

[Friederike]

How was the alignment done (has there been any alignment done?)?

[pbigbig]

Thank you, I used kmer-based searching for target sequence.

[Friederike]

Did you extract reads containing certain k-mers from a FASTQ file? If you don't have any shared coordinate space among the different reads then you either need to go via assembly or alignment to line and pile up similar reads

[pbigbig]

Yes, I did extracted reads using certain k-mer patterns.

Answer 1

[GenoMax]

You could try doing a local assembly using something like tadpole.sh from BBMap suite. A guide is available. You could do a multiple sequence alignment after converting the reads to fasta. Depending on how many you have this may or may not work well.

[pbigbig]

Thank you for your suggestion, I will try this.