Cancer Type Classification in Liquid Biopsies Based on Sparse Mutational Profiles Enabled through Data Augmentation and Integration

Multiple-instance learning of somatic mutations for the classification of tumour type and the prediction of microsatellite status

Large-scale genomic data are well suited to analysis by deep learning algorithms. However, for many genomic datasets, labels are at the level of the sample rather than for individual genomic measures. Machine learning models leveraging these datasets generate predictions by using statically encoded measures that are then aggregated at the sample level. Here we show that a single weakly supervised end-to-end multiple-instance-learning model with multi-headed attention can be trained to encode and aggregate the local sequence context or genomic position of somatic mutations, hence allowing for the modelling of the importance of individual measures for sample-level classification and thus providing enhanced explainability. The model solves synthetic tasks that conventional models fail at, and achieves best-in-class performance for the classification of tumour type and for predicting microsatellite status. By improving the performance of tasks that require aggregate information from genomic datasets, multiple-instance deep learning may generate biological insight.

Polyphenol Mediated Suppression of Hepatocellular Carcinoma (HepG2) Cell Proliferation by Clerodendrum infortunatum L. Root

OBJECTIVE

Clerodendrum infortunatum L. has long been used in traditional medicine in Sri Lanka for tumours, cancer, and certain skin diseases. The present study aimed to assess the anticancer properties of the aqueous extract of C. infortunatum L. root (AECIR) through the activation of the apoptotic pathway on hepatocellular carcinoma (HepG2) and thus give it a scientific validation. Further, the contribution of polyphenols in antioxidant activity and cell cytotoxicity was investigated.

METHODS

Powdered plant material was boiled with water (100°C) to obtained AECIR.  The DPPH assay was used to determine the antioxidant potential. The activity of AECIR on HepG2 and normal rat fibroblast (CC1) cell growth was determined using MTT assay. The morphological changes related to apoptotic pathway was examined by Ethidium Bromide/Acridine Orange (EB/AO), Rhodamine 123 (Rh123) and DNA fragmentation assay.

RESULTS

The AECIR demonstrated antioxidant potential with an EC50 of 350.2 ± 1.5 ug/mL for DPPH assay. The HO•, H2O2 and •NO free radical scavenging activity was observed with EC50 of 19.7 ± 2.3, 11.7 ± 0.1 and 273.1 ± 0.9 ug/mL, respectively. The antiproliferative effect of AECIR on HepG2 cells was observed in a time and dose dependent manner with an EC50 of 239.1 ± 1.3 μg/mL while CC1 cells showed a nontoxic effect with an EC50 1062.7 ± 3.4 μg/mL after 24hrs treatment. A significant decrease in antioxidant activity (p<0.001) and 90% HepG2 cell viability was observed with polyphenol removed AECIR compared to the polyphenol present AECIR. The EB/AO uptake, depletion of mitochondrial transmembrane potential, and DNA fragmentation assay results revealed that the apoptosis was induced by AECIR.

CONCLUSION

The obtained result of the present study demonstrates that the antioxidant potential and antiproliferative activity of AECIR is attributed to the presence of polyphenols. Furthermore, the findings provide the scientific base for anti-cancer potential of AECIR.

Mutation-Attention (MuAt): deep representation learning of somatic mutations for tumour typing and subtyping

BACKGROUND

Cancer genome sequencing enables accurate classification of tumours and tumour subtypes. However, prediction performance is still limited using exome-only sequencing and for tumour types with low somatic mutation burden such as many paediatric tumours. Moreover, the ability to leverage deep representation learning in discovery of tumour entities remains unknown.

METHODS

We introduce here Mutation-Attention (MuAt), a deep neural network to learn representations of simple and complex somatic alterations for prediction of tumour types and subtypes. In contrast to many previous methods, MuAt utilizes the attention mechanism on individual mutations instead of aggregated mutation counts.

RESULTS

We trained MuAt models on 2587 whole cancer genomes (24 tumour types) from the Pan-Cancer Analysis of Whole Genomes (PCAWG) and 7352 cancer exomes (20 types) from the Cancer Genome Atlas (TCGA). MuAt achieved prediction accuracy of 89% for whole genomes and 64% for whole exomes, and a top-5 accuracy of 97% and 90%, respectively. MuAt models were found to be well-calibrated and perform well in three independent whole cancer genome cohorts with 10,361 tumours in total. We show MuAt to be able to learn clinically and biologically relevant tumour entities including acral melanoma, SHH-activated medulloblastoma, SPOP-associated prostate cancer, microsatellite instability, POLE proofreading deficiency, and MUTYH-associated pancreatic endocrine tumours without these tumour subtypes and subgroups being provided as training labels. Finally, scrunity of MuAt attention matrices revealed both ubiquitous and tumour-type specific patterns of simple and complex somatic mutations.

CONCLUSIONS

Integrated representations of somatic alterations learnt by MuAt were able to accurately identify histological tumour types and identify tumour entities, with potential to impact precision cancer medicine.