Option to only do one side

[mbhall88]

I would like to only extend one side of my linear contig. For instance, I am working with linear plasmids in E. faecium and they usually have a hairpin end on the left, and an open right end. I'd like to only extend the right end.

[dalofa]

I had not thought about this being a relevant option but would be more than willing to implement it.

I am very excited about hearing from someone that works with linear replicons not from streptomycetes! Is it common that the right end/non-hairpin end is recessed in assemblies of E. faecium?

[mbhall88]

Well to be honest I hadn’t given it a huge amount of attention until I saw your preprint. Ryan Wick was helping me with resolving the open end of some of these plasmids and there definitely seems to be some variability in what different assemblers give you. One example I ran telomere on found and extra few hundred bases which is promising

[tuspjo]

I would love to look into the telomore results from this, if you are ok sharing them? It's extremely interesting for us to see if the principles which we've spent a lot of time validating in Streptomyces also works well in non-streptomycetes which might have other linear end issues.

For your question to only look at one end, Telomore run on both ends should extend only one if the other end is already complete! if it behaves differently, that's a github issue!

Cheers

[mbhall88]

For your question to only look at one end, Telomore run on both ends should extend only one if the other end is already complete! if it behaves differently, that's a github issue!

Well the hairpin end kind of breaks this assumption. Because when a real extends betond the hairpin end, it is just extending back on itself. For example, here is an IGV screenshot from the QC BAM provided by telomore

You can see that all alignments are essentially centred on the same point (the original left end of the contig). The alignments are sorted by read name and coloured by orientation. So each read basically aligns twice around this point on both strands as they are reads that loop back around on the hairpin.

When I use the option to keep all intermediate files I see there are the following files relating to the right side

LRE_04_right.fasta
LRE_04_right.sam
LRE_04_right_cons.fasta
LRE_04_right_cons.fasta.aln
LRE_04_right_filtered.fastq
LRE_04_right_filtered.sam
LRE_04_right_map.bam
LRE_04_right_map.bam.bai
LRE_04_trunc_right.fa

which one of these is added to the original contig to extend it? i.e., if I want to create a right-only extended version of my contig, which of these sequences is the right one? or is it not as simple as that?

[dalofa]

Hi

This is really fascinating! It would be interesting in the future to write some logic to resolve the hairpins as well. I think this is already implemented in Autocycler, however.

I have implemented the option to skip a side on the branch "skip_side" and am currently testing it. Either way this information might be useful if people run telomore with --keep. I have reorganized the files in the order they are created and they stage in which they are created:

Mapping Reads
LRE_04_right.sam This is all terminal reads mapping at the "right " terminus of the replicon (within the last 20 bases).
LRE_04_right_filtered.sam Terminal reads that extend beyond the reference at the right-side.
LRE_04_right_filtered.fastq All terminal reads that extend beyond the reference at the right-side. Includes only the soft-clip part of the read + (up to) 500 bases that "anchor" the read to the replicon.

Generating Consensus
LRE_04_right_cons.fasta.aln Alignment file produced during consensus building by lamassemble/mafft.
LRE_04_right_cons.fasta Consensus produced from LRE_04_filtered.fastq by lamassemble/mafft. Thus, it is a consensus of the soft-clipped part of reads + anchor bases.

Extending Assembly
LRE_04_trunc_right.fa Right-side of the replicon after splitting it in two. Used for mapping the consensus+anchor bases since Streptomycetes often have terminal inverted repeats.
LRE_04_right_map.bam Consensus+anchor bases mapped onto a truncated reference (LRE_04_trunc_right.fa). Used to generate LRE_04_right.fasta and the extended replicon later.
LRE_04_right_map.bam.bai Corresponding index-file
LRE_04_right.fasta This is the consensus/sequence that the replicon is extended with. It is the soft-clipped part of the consensus+anchor bases mapped to the truncated right side of the replicon. This is the sequence you are looking for if you want to extend the genome manually by appending sequence. This sequence has been trimmed.

Hope this helps!

[tuspjo]

Ahh, I see @mbhall88, yes, you're right, that does break the logic. And as you suggest, a fix would be to limit telomore to one end. (a harder fix would identify near perfect inverted repeat reads and use them to determine the hairpin position, but I think that's outside of the scope of telomore).
What a cool example, so the molecule is effectively one big inverted repeat and linear, and library prep will break both strands so reads which loop around the hairpin are symmetrical.