Naslov
APOBEC mutational signature affects prediction of cell of origin in different molecular groups of breast cancer

Autori
Štancl, Paula ; Karlić, Rosa ; Polak, Paz

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
PhD student symposium 2021 / Barišić, Dajana - , 2021, 249-250

Skup
Simpozij studenata doktorskih studija PMF-a

Mjesto i datum
Zagreb, Hrvatska, 24.04.2021. - 25.04.2021

Vrsta sudjelovanja
Poster

Vrsta recenzije
Podatak o recenziji nije dostupan

Ključne riječi
ishodišna stanica tumora, mutacijski potpisi, APOBEC
(cell-of-origin, mutational signatures, APOBEC)

Sažetak
One of the main challenges in cancer biology is the identification of the cell-of-origin (COO) of cancers. It has been shown that cell type of origin of a cancer can be accurately determined based on the distribution of mutations along its genome [1]. Novel approaches based on machine learning methods can successfully identify COO by utilizing different epigenetic features of the COO and somatic mutations, since the distribution of mutations reflects the chromatin structure of the cell-of-origin. However, these models can not explain well the variance in somatic mutations across the genome in certain cancers, such as breast cancer [2]. Often in human cancers, there is a biological noise in the genome due to hypermutation phenomena and specific mutational processes, such as those due to the APOBEC family of cytidine deaminases [3], which is especially present in breast cancers. This noise can negatively affect the outcome of predicting the cell-of-origin in cancers. The goal of this research is to explore the impact of genomic regions with high APOBEC mutational signature on the outcome of the models for predicting the cell- of-origin in different subtypes of breast cancer (basal and non-basal breast cancers). Mutational signatures are characteristic patterns of somatic mutations caused by distinct mutational processes. Random forest regression was used to model the distribution of somatic mutations in 1 Mb windows using the chromatin profiles from various normal tissues. Since mutations that are due to APOBEC activity may lower the prediction accuracy of the model, the modelling was performed with removed regions containing the APOBEC signature above a certain threshold. For each subtype of breast cancer, the best cutoff value of percentage of APOBEC signature for a region was determined based on the performance of the model. Both basal-like breast tumors and non-basal-like tumors had an increase in overall explained variance when removing regions with a high percentage of APOBEC signature. However, the best cut-off values for subtypes vary, for basal breast tumors the cut-off value is 10% while for non-basal tumors it is about 40%. Basal-like breast tumors have significantly smaller amounts of APOBEC signature compared to non-basal-like tumors therefore this difference in cutoff values is expected. Analysis of genomic regions with different proportions of APOBEC signature indicates that the mutations that result from APOBEC activity, primarily occur in open chromatin regions. Together these results indicate the importance of characteristics of regions and mutations which can be used to improve the prediction of cell-of origin in cancers.

Izvorni jezik
Engleski

Znanstvena područja
Biologija

POVEZANOST RADA