Fam3PRO Update Log

Fam3PRO 2.0.4

This release includes several important updates and improvements:

• Updated VizRisk_singleFutureRisk() which uses the plotly package for fully interactive plots.
• Enhancements to singleFutureRisk()Added support for custom race input.
• Improved internal logic to ensure that cancer sites with zero risk for a given sex (e.g., male ovarian cancer) are properly excluded from results.

Fam3PRO 2.0.3

This release includes several important updates and improvements:

• The Fam3PRODatabase was updated to reflect recent gene-cancer association changes, including manual adjustments to penetrance values for gene-cancer pairs without established associations.
• Updated a convenient plotting function to visualize individual cumulative future cancer risks vs general population risks.
• Fix variant filtering in buildDatabase to retain CHEK2 and NBN.

Fam3PRO 2.0.2

This release adds new functionality for single-proband future cancer risk estimation and visualization.

• Added singleFutureRisk() function: a simplified wrapper to compute future cancer risks for probands with a single pathogenic variant, conditioning on current age and sex.
• Added VizRisk_singleFutureRisk() function: generates a clean visualization of future risk trajectories over age.
• Improved documentation and examples for both functions.

Fam3PRO 2.0.1

This release fixes some bug.

• Fixed age imputation bug where upper and lower bounds were not correctly swapped.
• Renamed PanPRO11 and PanPRO22 to Fam3PRO11 and Fam3PRO22.
• Added example with BCRAT variables to the help file.
• Fixed installation issue where cbcrisk was not automatically installed with Fam3PRO.

Fam3PRO 2.0.0

This release includes several important changes and improvements:

• Major changes

1. Introduction of use.mult.variant Parameter
A new parameter, use.mult.variant, has been introduced to improve precision in cancer risk prediction for families with multiple variants in the BRCA1 and BRCA2 genes. Previously, the model only used general markers:

BRCA1_hetero_anyPV
BRCA2_hetero_anyPV
Now, with use.mult.variant = TRUE, the model can specify and differentiate between:

BRCA1:
BRCA1_hetero_BCCR (Breast Cancer Cluster Region)
BRCA1_hetero_OCCR (Ovarian Cancer Cluster Region)
BRCA1_hetero_other
BRCA2:
BRCA2_hetero_BCCR (Breast Cancer Cluster Region)
BRCA2_hetero_OCCR (Ovarian Cancer Cluster Region)
BRCA2_hetero_other
This feature enhances the accuracy of predictions by taking into account specific variants within these genes, which is particularly useful for families with multiple pathogenic variants.

Examples of how to use use.mult.variant have been added to the GitHub documentation.

2. Introduction of breakloop Parameter
To handle complex pedigrees with loops, we have introduced a new parameter, breakloop:

When breakloop = TRUE, the function automatically detects loops in the pedigree and “breaks” them by creating clone entries where necessary.
This ensures that pedigrees with loops can be processed without causing infinite recursion or calculation errors.
Detailed examples of how to use breakloop have been added to the README and package documentation.

3. Update of SEER Penetrance Data
The package now incorporates the latest SEER data to update cancer penetrance estimates. This ensures that our risk models are aligned with the most current population-level cancer statistics, thereby improving predictive accuracy.

4. Rename of Package and Core Functions
The package has been renamed from PanelPRO to Fam3PRO. This change includes:

Renaming core functions, documentation, and internal references.
The new name aims to better align with the package’s expanded focus beyond just cancer panels.

5.Removal of Specific Models
The following models have been entirely removed from the package to simplify the structure and reduce the complexity of maintaining legacy models:

BRCAPRO
BRCAPRO5
BRCAPRO6
MMRPro

• Minor changes

6. Documentation Enhancements
Updated the README and vignettes to reflect the new package name and the changes in functionality.
Added detailed explanations and examples for the use.mult.variant and breakloop features.
Improved help files for core functions to ensure users have clear guidance on implementing the new features.

7. Code Optimization and Bug Fixes
Updated check files.

PanelPRO 1.1.0

This release includes several important changes and improvements:

• Manual installation of the cbcrisk v2.0 package is no longer required when installing PanelPRO. Instead, the first time the user runs the PanelPRO() function (or one of its model specific wrapper functions such as BRCAPRO()), cbcrisk will be automatically installed from GitHub, if it is not already.
• The intervention columns riskmod and interAge, which stored information on the prophylactic surgeries (mastectomies, oophorectomies, and hysterectomies) in list type format, made sharing .csv files of PanelPRO pedigrees difficult due to the fact that .csv files cannot easily handle list type columns. Therefore, PanelPRO now also accepts pedigrees with six different numeric columns to replace the riskmod and interAge list type columns. The naming conventions for the six new columns are riskmodX (3 columns) and interAgeX (3 columns), where “X” is one of the 3 abbreviated surgery names "Mast", "Ooph", or "Hyst". See the README for information on how to encode these columns.

PanelPRO 1.0.0

This release includes several important changes and improvements:

• Contralateral breast cancer (CBC) has been added to PanelPRO 1.0.0. This requires the user to install the non-CRAN R package cbcrisk manually, prior to installing PanelPRO 1.0.0. Due to the new dependency, if a user updates to PanelPRO 1.0.0, their calls to PanelPRO functions may result in errors if they do not install cbcrisk. Additional and optional columns related to breast cancer are now accepted in the input pedigree to support the addition of CBC. See the README file for information on how to install cbcrisk and the additional columns.
• Relatives in the input pedigree who are disconnected are now removed from the pedigree before analysis and a warning message informs the user of which relatives were disconnected.
• A small but fatal bug for the edge case where somebody wants to get net future risk for the proband’s CurAge+1. This bug unintentionally dropped dimensions in the cancer penetrance when subsetting for a single age like that results in some mischief.

PanelPRO 0.3.0

This release includes several important changes and improvements:

• APC, BMPR1A, and cervical cancer are not supported and have been removed from PanelPRODatabase. PanelPRO-22 replaces PanelPRO-24 as the largest model.
• Our age imputation procedure imputes missing current ages, followed by missing cancer affectation ages. If a missing cancer age cannot be sampled such that it satisfies its age bounds and the upper bound set by the current age, this imputation run will be discarded and re-attempted, with a warning message issued. If the desired number of runs (iterations, default 20) cannot be imputed within max.iter.tries (default 5*iterations) attempts, an error will be raised.
• Missing current ages are now imputed based on age distributions from the literature, instead of truncated normal distributions.
• When any of the allele frequencies for an ancestry other than "nonAJ" differ from the "nonAJ" allele frequencies (for a given model specification), the noncarrier penetrances for this ancestry gets set to the SEER penetrances for each cancer in the built database.
• Documentation notes that Asian-Pacific Islanders should be classified as "Asian" in the race column of the user pedigree.
• Ages between 0 and 1 will be rounded up to 1; other decimal ages will be rounded to the nearest integer. In both cases, an AgeRounding message will be printed.
• The visualization function visRisk now has an option to return the resulting Plotly object instead of plotting it. Checks for probands without carrier probability or future risk estimates (due to the model not specifying genes or cancers) are also implemented.
• A random.seed can now be specified for checkFam, such that the age imputations will match those in PanelPRO when the same random seed is used.
• New internal function checkGeneCancerAssociations automates checking for and visualizing gene-cancer associations.
• Master function documentation clarifies that ancestry information is used to select allele frequencies for BRCA1 and BRCA2, specifically.
• The use.mult.variants argument has been removed from the master function, since we currently do not support multiple variants. The documentation also notes that we output both heterozygous and heterozygous carrier probabilities when the database information is available (currently only MUTYH; references for this added).
• The checkMating checkFam check that none of a given individual’s relatives overlap with the individual’s spouse’s relatives has been expanded to include more spouses (i.e. relatives by marriage) in the two relative groups. This still doesn’t cover all possible loop scenarios, but it helps catch more of them.
• Removed the deprecated normalize option from the peeling-paring function, since we now always normalize the carrier probabilities and do not allow the user to specify otherwise.
• Added clarification (with examples) on how inheritance probabilities should be interpreted to the .calInheritanceProb documentation.
• When ages were imputed, warning messages will be printed that give each proband’s lower and upper bounds for the probability of carrying any pathogenic variant and encourage the user to provide more age information when possible.
• Fixed buildDatabase error that occurs when no cancers are passed in and the function attempts to coerce the (empty) cancers vector to title-case.
• Fixed ageImpute bugs involving inconsistent encoding of the Sex variable and relative information slots.
• If the lower and upper age bounds get swapped during ageImpute (i.e. because the lower bound was greater than the upper bound), the new bounds are now ensured to be valid (i.e. the new upper bound must be below MAXAGE).
• checkFam now raises an error when a MotherID or FatherID in the pedigree has an invalid sex.
• Interpretation of biomarker testing information in the documentation for PanelPRODatabase has been corrected; they should be the probabilities of the marker given genotype, not the other way around.
• checkFam check that drops biomarker testing for those unaffected for the relevant cancer has been fixed.
• Bug in cancer penetrance for ovarian cancer and BRCA2 has been fixed.
• Fixed some small typos in documentation and code comments.

PanelPRO 0.2.0

• The PanelPRO-24 model specification has been updated to include 18 instead of 11 cancers: Brain, Breast, Cervical, Colorectal, Endometrial, Gastric, Hepatobiliary, Kidney, Leukemia, Melanoma, Ovarian, Osteosarcoma, Pancreas, Prostate, Small Intestine, Soft Tissue Sarcoma, Thyroid, and Urinary Bladder.
• Eight additional test pedigrees have been added: test_fam_5 through test_fam_12.
• PanelPRODatabase has been updated to remove the association between male breast cancer and BRCA1.
• When an individual in the pedigree has an age of cancer diagnosis that is greater than their current age, checkFam now raises an error and suggests that if the user cannot resolve the age inputs, the maximum cancer age can be entered as the current age.
• When one or more of the carrier probabilities is 0, germline testing results were incorporated, and the default sensitivities/specifities of 1 were used, a warning will be issued that suggests a more accurate estimate may be achieved with user-specified sensitivities and specificities.
• The new ignore.proband.germ argument in PanelPRO allows users the option to ignore proband germline testing results (if provided) from being considered by the model.
• When the current age of one or more living probands is equal to PanelPRO:::MAXAGE, no future risk probabilities will be calculated for these individuals.
• A bug that scrambled the model parameters for the breast tumor markers has been fixed.
• Imputed ages can no longer exceed the legal range of ages (between 1 and PanelPRO:::MAXAGE).