We finally uploaded a preprint on Locityper—targeted genotyper for complex genes, on which @tobiasmarschal and I worked for almost two years!.

RT @MishaKolmogorov: So excited to have this preprint out! Severus is a new tool for somatic SV calling in cancer using long reads. It supp….

And thank you to our co-authors Elizabeth Plender, Guiscard Seebohm, Sven Meuth and @EichlerLab.

Locityper carefully manages read alignments, insert sizes and read depth profiles to identify the two most likely locus haplotypes (gene alleles) for a given short or long-read WGS dataset.

Parascopy is available on github ( and can be installed with conda (conda install -c bioconda parascopy). Thank you for reading! Please check the paper for more details [7/7].

To fix that, we run EM-algorithm to jointly find a subset of reliable PSVs and paralog-specific copy numbers for all samples. We require many samples to find the set of reliable PSVs. However, one can detect CNVs for a single sample if the region was analyzed previously. [6/7]

We use paralogous sequence variants (PSVs) — small differences between repeat copies — to find paralog-specific CN. Unfortunately, many PSVs are unreliable (same allele on several repeat copies), the fact that is often ignored when analyzing duplicated regions. [5/7].

To find aggregate CN, we use a Hidden Markov Model. Observations: read depth in consecutive 100 bp windows. Hidden states: aggregate CN values. We refine model parameters based on multiple samples to find deletions and duplications more accurately. [4/7]

As input, Parascopy takes single or multiple short-read WGS BAM/CRAM files and a list of target low-copy repeats. For each region, Parascopy outputs aggregate copy number (total CN across all repeat copies) and paralog-specific CN (CN for each individual repeat copy). [3/7]

Low-copy repeats, also known as segmental duplications, cover up to 5% of the genome, are prone to extensive copy number variation and are notoriously hard to analyze. In this paper, we present a novel computational method, Parascopy [2/7].

I am delighted to share our new paper in @NatureComms on identifying copy number for duplicated genes! #bioinformatics #genomics #structuralvariation. A short thread below: [1/7].

RT @HiTSeq: Copy number discussions continue with @TimofeyProdanov: "Robust and accurate estimation of paralog-specific copy number for dup….

Check out our paper on mapping long reads to segmental duplications: