Generalforsamling 2023

Tid og Sted: 17:15, Auditorium 10, H.C. Ørsted Instituttet, Universitetsparken 5, 2100 København Ø. Generalforsamlingen foregår på dansk. Efterfølgende vil der være foredrag ved Marta Pelizzola (se detaljer nedenfor)

Efter generalforsamlingen vil der være middag på restaurant i nærheden. Tilmelding kan foretages ved at skrive til formanden (chair@dsts.dk). Deadline for tilmelding 20. Februar. Middag forventes at koste 250 kr excl. drikkevarer.

Dagsorden

1. Valg af dirigent.
2. Bestyrelsens beretning fremlægges til godkendelse.
   • Udvalget for Presse og Formidling fortæller om udvalgets aktiviteter.
   • Young Statisticians Denmark fortæller om sine aktiviteter.
3. Regnskabet fremlægges til godkendelse.
4. Fastsættelse af næste års kontingent.
   Bestyrelsen foreslår at kontingentet fastholdes uændret:
   Studerende: 0 kr. Ordinære medlemmer: 100 kr. Pensionister: 50 kr.
5. Udvalget for Presse og Formidling fortæller om udvalgets aktiviteter.
6. Ny hjemmeside: Evaluering og input fra generalforsamlingen.
7. Bestyrelsen foreslår at anonymiserede indskannede meddelelser i vores nye elektroniske arkiv skal være offentligt tilgængelige.
8. Behandling af fremsendte forslag.
   Bestyrelsen har ikke modtaget nogle forslag
9. Valg af bestyrelsesmedlemmer.
   Christian Pipper træder ud af bestyrelsen. Bestyrelsen indstiller Birgitte Rønn (Zealand Pharma A/S) som kandidat
10. Valg af revisor.
    Bestyrelsen foreslår Anders Rønn-Nielsen (CBS-Finance).
11. Eventuelt.

Talk by Marta Pelizzola (Institut for Matematik, Aarhus Universitet)

Title: Robust learning of mutational signatures using non-negative matrix factorization in large nucleotide contexts

Abstract: Mutational signatures describe the pattern of mutations over the different mutation types. Usually, mutation types are defined by the base mutation and the flanking nucleotides at the right and left of the base mutation. We extend this definition and include flanking nucleotides further away from the base substitution. The importance of larger contexts lies in the way they affect mutation rates. Furthermore, when looking at more flanking nucleotides, it becomes relevant to consider mutational opportunities for different mutation types. Indeed, sets of nucleotide sequences occur with various rates along the genome, and this results in very different opportunities for the mutation types to occur.

Mutational signatures are usually derived using non-negative matrix factorization (NMF) in tri-nucleotide contexts. To extract the mutational signatures we have to choose an error model for the observed mutational counts, which determines the underlying distributional assumption of the data. In most applications, the mutational counts are assumed to be Poisson distributed, but this is often overdispersed and leads to an overestimation of the number of signatures. We modify the current model including parameterized signatures combined with mutational opportunities and we show that these provide more robust and interpretable signatures. Using cross-validation, we also show that including opportunities increases the predictive power of the model on new data. Overall, the use of mutational opportunities and parameterized signatures provides more robust signatures and a more accurate representation of the underlying mutational processes.