Teloclip¶

Recover unassembled telomeres from raw long-reads using soft-clipped read alignments.

The problem¶

Genome assemblers are conservative at chromosome ends. Telomeric repeats are short, highly repetitive and error-prone to sequence, so an assembler will usually stop before it reaches the true end of a chromosome, leaving the telomere out of the assembly.

The information is often still in your reads. A long read that starts inside the assembled contig and continues past its end will align to the contig for part of its length and then stop, with the remainder reported as a soft clip. That clipped tail is sequence the assembler declined to place.

contig  ==========================================>|
read 1        ---------------------------->
read 2               ------------------------------>~~~~~~~~
read 3                    ------------------------->~~~~~~~~~~~~
                                          aligned  |  soft-clipped

If several reads are clipped at the same contig end and their clipped tails contain telomeric repeats, that is good evidence the telomere is real and simply missing from the assembly.

Teloclip finds those reads, and can splice their clipped tails onto the contig.

The workflow¶

flowchart LR
    A[Long reads] --> B[minimap2]
    R[Draft assembly] --> B
    B --> C[teloclip filter]
    C --> D[teloclip extract]
    C --> E[teloclip extend]
    D --> F[Manual inspection]
    E --> G[Extended assembly]
    G --> H[Polish]

Command	What it does
`filter`	Keeps only alignments soft-clipped at a contig end, optionally requiring telomeric motifs
`extract`	Writes the overhang reads to per-contig-end FASTA files for inspection
`extend`	Splices the best overhang onto each contig end and writes a new assembly

Where to start¶

New to teloclip? Work through the tutorial.
Want to understand what counts as an overhang? See How overhangs are defined.
Results look wrong? Start with Biological cases that cause trouble.

Extended contigs are not polished

extend splices in sequence from a single raw read. Long reads carry indel errors, so the extension inherits them. Always re-polish the extended assembly with your long and short read data before downstream analysis.

Citing¶

If teloclip is useful in your work, please cite it. See CITATION.cff.