Cristian Groza, PhD
cgroza.bsky.social
Cristian Groza, PhD
@cgroza.bsky.social
That's why I always start conversations by guessing higher. It's more flattering to be overestimated than underestimated.
May 2, 2025 at 10:06 PM
Who would listen to AI generated music?
April 13, 2025 at 3:57 PM
This is what happens when you don't silence your transportable elements.
November 18, 2024 at 6:17 PM
In conclusion, we suggest a way to use pangenomes to call SVs against a reference pangenome like HPRC to filter out common SVs and find the very rare SVs that are unique to a disease genome. This allows clinicians to focus on a set of SVs that are more likely to be pathogenic.
January 22, 2024 at 6:15 PM
Therefore, we created a PBSV and minigraph consensus to obtain consensus rare SVs. When used Phrank to rank the consensus rare SVs, we uncovered a deletion in KMT2E, which is likely to be a causal variant for a previously undiagnosed case.
January 22, 2024 at 6:15 PM
When looking for the very rare alleles, we found that current SV methods such as PBSV and minigraph generate too many false positive calls for a clinician to curate in a reasonable amount of time. This is especially so since false positive show up as rare SVs.
January 22, 2024 at 6:14 PM
We found that minigraph calls 29,964 in the twins, of which 84.96% are in both twins. PBSV calls 23,516 SVs in the same twins, of which 83.12% are in both twins.
January 22, 2024 at 6:13 PM
We found many alleles, but it would be helpful to know how assembly and pangenome approaches compare with long read SV callers such as PBSV. We made use of a pair of twins that is present within GA4K, and checked how many SVs are called in both twins using minigraph vs PBSV.
January 22, 2024 at 6:12 PM
Overall, we have found 204,551 SV alleles that are unique to the GA4K cohort, with the most common allele occurring in only 88 of the 574 haplotypes. This highlights the importance of personal de novo assemblies in finding rare SVs.
January 22, 2024 at 6:11 PM
The additional sequence is mostly composed of simple repeats, satellites, and TEs. However, we estimate that at least 4.7 Mbp is unique sequence that is not repeats and is not present in HPRC or in the reference genome.
January 22, 2024 at 6:10 PM
First, we assembled 574 rare disease haplotypes using HiFi, hifiasm and parental data. We validated these assemblies with Flagger developed within HPRC. Then we built a pangenome graph with minigraph on top of HPRC. This added 426 Mbps that is specific to the GA4K cohort.
January 22, 2024 at 6:09 PM
It's seasoning, just like well kept cast iron.
January 20, 2024 at 6:20 PM
MPs being mad about the tax hike reveals who they represent.
January 17, 2024 at 8:12 PM
2FA on top of ssh keys 🤕
January 2, 2024 at 4:04 PM