<i>Popfinder</i>: A Highly Effective Artificial Neural Network Package for Genetic Population Assignment
Bibliographic record
Abstract
The ability to assign biological samples to source populations with high accuracy and precision based on genetic variation is important for numerous applications from ecological studies through wildlife conservation to epidemiology. However, population assignment when genetic differentiation is low is challenging, and methods to address this problem are lacking. The application of artificial neural networks to population assignment using genomic data is highly promising. Here we present popfinder: a new, easy-to-use Python-based artificial neural network pipeline for genetic population assignment. We tested popfinder both with simulated genetic data from populations connected by varying levels of gene flow and with reduced-representation sequence data for three species of seabirds with weak to no population genetic structure. Popfinder was able to assign individuals to their source populations with high accuracy, precision and recall in most cases, including both simulated and empirical data sets, except in the empirical data set with the weakest population structure, where the comparator programs also performed poorly. Compared to other available software, popfinder was slower on the simulated data sets due to hyperparameter tuning and the fact that it does not reduce the dimensionality of the data set; however, all programs ran in seconds on empirical data sets. Additionally, popfinder provides a perturbation ranking method to help develop optimised SNP panels for genetic population assignment and is designed to be user-friendly. Finally, we caution users of all assignment programs to watch both for leakage of data during model training, which can lead to overfitting and inflation of performance metrics, and for unequal detection probabilities.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.010 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.004 | 0.003 |
| Research integrity | 0.001 | 0.004 |
| Insufficient payload (model declined to judge) | 0.065 | 0.023 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".