CAGI Gathering at ASHG

We will be holding an informal CAGI gathering at ASHG in Vancouver on either Wednesday, 19 October from 6pm-7pm or Thursday, 20 October from 3:30pm-4:30pm.

Please fill in the Doodle pool:

to help us select a suitable time. The Wednesday time partially overlaps with an Award presentation and the Thursday time with the poster session. We will confirm the location at a closer time.

Please write us at to meet with us if you cannot make these times.

There will also be a CAGI poster presented at the ASHG conference in Vancouver. Come and meet Gaia on Wednesday, 3:00-4:00 pm, at poster 1822 which has material in progress for the flagship paper.

We are currently looking for new data sets and new data providers for CAGI 5, and people we could meet in Vancouver. If you have ideas for contacts or recommendations for CAGI challenge datasets, we would much appreciate hearing about them. We have a few leads on challenges, including those that provide continuity from the previous experiments.

* CAGI special issue

We would like to thank all assessors and predictors who submitted their manuscripts. Of these,10 papers are currently under review. We expect to have a total of 23 papers for this special issue.

* CAGI flaghship

We continue to develop the flagship manuscript and welcome further input expanding the 15 themes that will be described in the manuscript, list below.

CAGI themes:

- Top missense prediction methods are highly statistically significant, but individual variant accuracy is limited

- Missense methods tend to correlate better with each other than with experiments

- Structure-based missense methods excel in a few cases, while evolution-based methods have more consistent performance

- Phenotype assignment based on rare variant analysis sometimes succeeds for a range of challenges

- Predicting complex traits from exomes is fraught

- Predictors were able to identify causal variants that were overlooked by a clinical laboratory

- Bespoke approaches often enhance performance

- Creating a good challenge that provides a reliable gold standard is difficult

- PGP ‘improvements’ (overall years)

- The clinical use of multiple uncalibrated missense methods is not advised

- Some challenges are just too hard for the current state of the art

- Interpretation of non-coding variants shows promise but is not at the level of missense

- The CAGI community has potential for addressing unanswered questions in genome interpretation

- CAGI challenges have contributing to interpretation of cancer variants

- There might be potential for missense interpretation at the extreme of the distribution

We hope to see you in Vancouver!

Best wishes,
Gaia, John, Steven