Madison Darmofal
YOU?
Author Swipe
View article: Ultrasensitive ctDNA monitoring for organ preservation in patients with locally advanced rectal cancer
Ultrasensitive ctDNA monitoring for organ preservation in patients with locally advanced rectal cancer Open
Optimal selection of patients with locally advanced rectal cancer for watch and wait (WW) and optimal management during follow-up remain challenging. We employed a primary-tumor-informed whole genome sequencing (WGS) assay to detect circul…
View article: Integrated histopathologic modeling of detailed tumor subtypes and actionable biomarkers
Integrated histopathologic modeling of detailed tumor subtypes and actionable biomarkers Open
Accurate cancer subtyping with accompanying molecular characterization is critical for precision oncology. While machine learning approaches have been applied to both digital pathology and cancer genomics, previous work has been limited in…
View article: RETRACTED: Circulating tumor DNA status to direct adjuvant immunotherapy for mismatch repair deficient tumors
RETRACTED: Circulating tumor DNA status to direct adjuvant immunotherapy for mismatch repair deficient tumors Open
21 March, 2025. This preprint was retracted at the authors' request due to a conflict with a conference embargo policy. Authors should verify embargo policies before posting to avoid post-publication removal.
View article: Table S12 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S12 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Top Shapley values per cancer type. Shapnorm = shapley effect score normalized across all predictions of the cancer type
View article: Table S16 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S16 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
GDD-ENS performance across tumor content thresholds
View article: Figure S9: Shapley Value Heatmaps Across Cancer Types from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S9: Shapley Value Heatmaps Across Cancer Types from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Heatmap of Association Scores across cancer types, defined as aggregated Shapley value effect scores across signatures (A), structural variants (B), and chromosome arms (C) with the highest p-value following Mann-Whitney U-test. For (C), r…
View article: Table S17 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S17 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
CUP Later confirmed Diagnoses
View article: Table S8 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S8 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Metrics for GDD-ENS performance across output confidence thresholds
View article: Table S7 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S7 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
All GDD-ENS test set results
View article: Table S3 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S3 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Cohort Demographic Table. TMB = Tumor Mutational Burden, SD = Standard Deviation
View article: Table S7 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S7 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
All GDD-ENS test set results
View article: Figure S6: Individual Type Shapley Values from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S6: Individual Type Shapley Values from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Top ten most important features leading to correct predictions of all 38 cancer types included, as approximated by normalized Shapley value effect scores. We sum absolute Shapley values per feature, indicate whether the feature was present…
View article: Table S4 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S4 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Stepwise accuracy of various models generated during model development, including direct comparison of GDD-RF to GDD-ENS. Acc. = Accuracy, Conf. = Confidence, Macro Prec: Class-Averaged Precision, % Excluded: proportion of high-content, so…
View article: Table S11A/B from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S11A/B from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
S11A: Expected Panel Performance (Masked Analysis), S11B: Performance of GDD-ENS on UCSF test set. Expected Acc: Recalibrated accuracy comparable to accuracy on GDD-ENS cohort after correction for difference in distribution of cancer types…
View article: Figure S1: Accuracy of Feature-Specific Classifiers from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S1: Accuracy of Feature-Specific Classifiers from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
For each broad category of features present within our training set, we trained individual models using the same training regime as GDD-ENS. Results are shown across these categories, represented by a circle. We then iteratively combined a…
View article: Table S6 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S6 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Accuracy across all individual MLP models that form GDD-ENS. FC = Fully Connected, ECE = Estimated Calibration Error
View article: Figure S5: GDD-ENS Performance Across Purity Values from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S5: GDD-ENS Performance Across Purity Values from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Assessment of GDD-ENS performance on the test set across reported purity values, binned in 10% increments. Count of each bin (top), and corresponding overall accuracy (light pink), and high-confidence accuracy (dark pink). Samples with NA …
View article: Table S13 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S13 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Cancer Type Organ System Mapping
View article: Figure S10: KRAS Shapley Values across typesSupplementary Data from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S10: KRAS Shapley Values across typesSupplementary Data from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Normalized absolute Shapley value scores for all KRAS-related features across cancer types with KRAS implicated within the top ten most predictive features per type. KRAS indicate presence of any broad alteration that affects KRAS within a…
View article: Figure S2: GDD-ENS Precision Recall Curves from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S2: GDD-ENS Precision Recall Curves from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Precision Recall curves for GDD-ENS test set performance across all individual cancer types (grey) and the average of all outputs (black).
View article: Table S16 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S16 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
GDD-ENS performance across tumor content thresholds
View article: Table S14 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S14 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Top Shapley values per organ system. Shapnorm = shapley effect score normalized across all predictions of the cancer type
View article: Figure S4: Ancestry Accuracy Differentials from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S4: Ancestry Accuracy Differentials from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
(A) Overall proportion of all annotated ancestries across the training set (left) and testing set (right). Numbers over each bar indicate total within each category. EUR, European; ADM, Admixed; EAS, East Asian; AFR, African; SAS, South As…
View article: Table S1 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S1 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
All Sample IDs and associated AACR-GENIE IDs, Cancer Types and classification category (train, test, CUP etc)
View article: Figure S11: Heatmap of Labels Mapped for Adaptable Prior Distributions from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S11: Heatmap of Labels Mapped for Adaptable Prior Distributions from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Proportion of all in-distribution discovery cohort samples that are either primary or metastatic (top), or have broad histological annotations (middle, top) per cancer type. Underlying distributions of metastatic site (middle) and histolog…
View article: Table S2 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S2 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Cancer Type and Detailed Cancer Type Labelling for Discovery Cohort Samples
View article: Figure S8: Organ Shapley Value Distributions from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Figure S8: Organ Shapley Value Distributions from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Shapley values aggregated across 10 major organ systems, as described in Supp. Table S13. First column represents broad feature category importance across all correct, in-distribution predictions. Second and third columns represent top fea…
View article: Table S3 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data
Table S3 from Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data Open
Cohort Demographic Table. TMB = Tumor Mutational Burden, SD = Standard Deviation