Refactor/refactor AlphaPeel by XingerTang · Pull Request #292 · AlphaGenes/AlphaPeel

XingerTang · 2026-07-07T17:57:24Z

Related Issue

Closes #78
Closes #147
Closes #172
Closes #178
Closes #258
Working...

What changed

Call modules of functional tests directly instead of building, installing, and running with an OS command
Check directly against the error message and warning message in the functional tests
Improve the readability of the functional test file
Use ValueError instead of printing error messages
Add instructions for functional tests memory profiling and coverage report generation
Simplify the accuracy test so that it only runs minimal
Provide separate full performance benchmarking in utility files.
Provide visualization of the full benchmarking report.

Working...

Bigger picture of the full plan

Modification on tests:
- For functional tests:
  - Call modules directly instead of build and install and run with OS command
  - It would enable:
    - memory profiling
    - function-wise runtime profiling
    - test coverage report
    - more informative debugging information when test failed
    - catch errors and warnings easier than the other way
- Keep the accuracy tests as the current build and install approach to test by platform and distribution
- Update accuracy test table as AlphaImpute2
Refactor the reassignment of argument options in AlphaPeel to remove warnings raised because of the unsynced arguments retrieval.
Remove unused/commented code
Replace single-letter variable names with more informative names
Add comments and docstrings when applicable
Memory optimisation
- Background: For a 2MB genotype input, the PeelingInfo would take 174MB
- By reviewing the current most memory-demanding data, several improvements can make:
  - PointSeg is used a temporary holder of segregation probabilities of one single individual, no necessity to allocate it an nInd x 4 x nLoci array.
  - Similar case but a little bit more complicated situation may apply to family terms:
    - posteriorSire_new
    - posteriorDam_new
    - posteriorSire_minusFam
    - posteriorDam_minusFam
  - There are several variables of int64 type, int64 is too large than what we need
Move the functions that are jit-compilable to jit
Explore multi-processing solution on HPC

Why this change

Refactoring the code

Working...

Submodule changes

Submodule repository: AlphaGenes/tinyhouse
PR: Refactor/refactor alphapeel tinyhouse#40
Why needed: argparsing, file I/O, ...

Notes / Risks

Working...

XingerTang · 2026-07-07T18:02:25Z

Coverage report of the functional test:

coverage report -m
Name                                            Stmts   Miss  Cover   Missing
-----------------------------------------------------------------------------
conftest.py                                       104     33    68%   10-18, 23-33, 58-71, 150-151, 198
src/tinypeel/Peeling/Peeling.py                   268    240    10%   31-243, 268-274, 292-303, 324-332, 352-360, 370-382, 412-427, 446-452, 471-477, 490-512, 525-542, 565-641
src/tinypeel/Peeling/PeelingIO.py                 173     68    61%   24-84, 151-152, 183, 191, 214, 267-285, 430-433, 447-450, 463-487
src/tinypeel/Peeling/PeelingInfo.py               211     74    65%   54, 153, 210-213, 232-235, 269, 289-300, 366-383, 391-434, 446-458, 470-482
src/tinypeel/Peeling/PeelingUpdates.py            191     74    61%   35, 54, 75, 77, 88, 90, 95, 113-135, 153-164, 261, 279, 336-344, 397-400, 413-437, 462-466, 503-509, 524-547, 560-564, 576
src/tinypeel/Peeling/version.py                     2      0   100%
src/tinypeel/tinyhouse/HaplotypeOperations.py      59     42    29%   6-8, 16-22, 31-39, 44-46, 51-54, 59-63, 68-73, 78-85
src/tinypeel/tinyhouse/InputOutput.py             343    156    55%   15, 42-62, 70-71, 76-77, 87-88, 98-106, 114-122, 243-258, 262-284, 308-309, 327-328, 331-340, 343, 357-358, 368-370, 415-416, 438-446, 453-454, 458-459, 481-482, 485-486, 510-567, 576-577, 590-604, 610-633, 639-642, 646-655, 659-666, 698, 700
src/tinypeel/tinyhouse/MultiThreadIO.py            78     44    44%   9-10, 15-16, 21-47, 52, 56-63, 97-105, 122-138, 151
src/tinypeel/tinyhouse/Pedigree.py                778    350    55%   45-48, 107, 110, 120-137, 143, 199, 212-214, 292, 300-310, 313-337, 342-363, 366, 402-408, 411-418, 422-427, 434, 437, 447-451, 483, 501-512, 530-557, 560-567, 665, 675, 677, 680-708, 722, 728, 732, 740, 767-789, 795-798, 805-824, 829-849, 855-872, 883-938, 981, 986-987, 1009-1021, 1091, 1097, 1104, 1116-1138, 1146, 1154-1159, 1176-1177, 1181, 1185-1186, 1193-1199, 1212, 1215-1226, 1229-1237, 1240-1244, 1247-1252, 1255-1264, 1270-1279, 1283-1294, 1298-1307, 1310, 1315-1317, 1322-1325, 1330-1332, 1337-1340, 1345-1348
src/tinypeel/tinyhouse/ProbMath.py                251    115    54%   35-55, 65, 67, 95-96, 106-140, 145-147, 152-164, 176-211, 215-254, 260-262, 267-270, 275-278, 283-289
src/tinypeel/tinypeel.py                          286     50    83%   97, 113-119, 146-149, 165-168, 300-327, 350-362, 812-813, 845, 849-855, 886-891, 899
tests/functional_tests/run_func_test.py           527      2    99%   79, 177
-----------------------------------------------------------------------------
TOTAL                                            3271   1248    62%

gregorgorjanc

The changes look good/ok

XingerTang · 2026-07-08T11:34:51Z

Functional tests memory profiling - this is not very informative, as the Numba function can be called at any part of the run:

======================================================================================= MEMRAY REPORT =======================================================================================
Allocation results for tests/functional_tests/run_func_test.py::TestClass::test_out_id_only at the high watermark

         📦 Total memory allocated: 90.3MiB
         📏 Total allocations: 899
         📊 Histogram of allocation sizes: | ▂█▂▁|
         🥇 Biggest allocating functions:
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 19.6MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 4.6MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 4.3MiB
                - _to_string:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/ir/values.py:549 -> 2.5MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 2.1MiB


Allocation results for tests/functional_tests/run_func_test.py::TestClass::test_files at the high watermark

         📦 Total memory allocated: 52.2MiB
         📏 Total allocations: 3811
         📊 Histogram of allocation sizes: |  █  |
         🥇 Biggest allocating functions:
                - _compile_bytecode:<frozen importlib._bootstrap_external>:729 -> 2.9MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 2.2MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 2.1MiB
                - _compile_bytecode:<frozen importlib._bootstrap_external>:729 -> 1.1MiB
                - updatecache:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/linecache.py:137 -> 1.1MiB


Allocation results for tests/functional_tests/run_func_test.py::TestClass::test_est at the high watermark

         📦 Total memory allocated: 9.4MiB
         📏 Total allocations: 482
         📊 Histogram of allocation sizes: | ▁█▄ |
         🥇 Biggest allocating functions:
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 1.2MiB
                - _icmp:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/ir/builder.py:575 -> 1.0MiB
                - init_first_block:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/numba/core/interpreter.py:1519 -> 1.0MiB
                - __init__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/ir/instructions.py:464 -> 1.0MiB
                - generic:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/numba/core/typing/npydecl.py:97 -> 1.0MiB


Allocation results for tests/functional_tests/run_func_test.py::TestClass::test_pheno at the high watermark

         📦 Total memory allocated: 9.0MiB
         📏 Total allocations: 397
         📊 Histogram of allocation sizes: |  █▁ |
         🥇 Biggest allocating functions:
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 3.4MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 2.0MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 604.1KiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 415.7KiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 170.1KiB


Allocation results for tests/functional_tests/run_func_test.py::TestClass::test_sex at the high watermark

         📦 Total memory allocated: 4.2MiB
         📏 Total allocations: 316
         📊 Histogram of allocation sizes: | ▁█▂ |
         🥇 Biggest allocating functions:
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 2.6MiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 185.3KiB
                - register:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/ir/_utils.py:21 -> 128.0KiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 77.2KiB
                - __call__:/Users/xtang3/anaconda3/envs/phase/lib/python3.11/site-packages/llvmlite/binding/ffi.py:197 -> 77.2KiB

XingerTang · 2026-07-08T11:42:41Z

Coverage report of the functional test:

coverage report -m
Name                                            Stmts   Miss  Cover   Missing
-----------------------------------------------------------------------------
conftest.py                                       104     33    68%   10-18, 23-33, 58-71, 150-151, 198
src/tinypeel/Peeling/Peeling.py                   268    240    10%   31-243, 268-274, 292-303, 324-332, 352-360, 370-382, 412-427, 446-452, 471-477, 490-512, 525-542, 565-641
src/tinypeel/Peeling/PeelingIO.py                 173     68    61%   24-84, 151-152, 183, 191, 214, 267-285, 430-433, 447-450, 463-487
src/tinypeel/Peeling/PeelingInfo.py               211     74    65%   54, 153, 210-213, 232-235, 269, 289-300, 366-383, 391-434, 446-458, 470-482
src/tinypeel/Peeling/PeelingUpdates.py            191     74    61%   35, 54, 75, 77, 88, 90, 95, 113-135, 153-164, 261, 279, 336-344, 397-400, 413-437, 462-466, 503-509, 524-547, 560-564, 576
src/tinypeel/Peeling/version.py                     2      0   100%
src/tinypeel/tinyhouse/HaplotypeOperations.py      59     42    29%   6-8, 16-22, 31-39, 44-46, 51-54, 59-63, 68-73, 78-85
src/tinypeel/tinyhouse/InputOutput.py             343    156    55%   15, 42-62, 70-71, 76-77, 87-88, 98-106, 114-122, 243-258, 262-284, 308-309, 327-328, 331-340, 343, 357-358, 368-370, 415-416, 438-446, 453-454, 458-459, 481-482, 485-486, 510-567, 576-577, 590-604, 610-633, 639-642, 646-655, 659-666, 698, 700
src/tinypeel/tinyhouse/MultiThreadIO.py            78     44    44%   9-10, 15-16, 21-47, 52, 56-63, 97-105, 122-138, 151
src/tinypeel/tinyhouse/Pedigree.py                778    350    55%   45-48, 107, 110, 120-137, 143, 199, 212-214, 292, 300-310, 313-337, 342-363, 366, 402-408, 411-418, 422-427, 434, 437, 447-451, 483, 501-512, 530-557, 560-567, 665, 675, 677, 680-708, 722, 728, 732, 740, 767-789, 795-798, 805-824, 829-849, 855-872, 883-938, 981, 986-987, 1009-1021, 1091, 1097, 1104, 1116-1138, 1146, 1154-1159, 1176-1177, 1181, 1185-1186, 1193-1199, 1212, 1215-1226, 1229-1237, 1240-1244, 1247-1252, 1255-1264, 1270-1279, 1283-1294, 1298-1307, 1310, 1315-1317, 1322-1325, 1330-1332, 1337-1340, 1345-1348
src/tinypeel/tinyhouse/ProbMath.py                251    115    54%   35-55, 65, 67, 95-96, 106-140, 145-147, 152-164, 176-211, 215-254, 260-262, 267-270, 275-278, 283-289
src/tinypeel/tinypeel.py                          286     50    83%   97, 113-119, 146-149, 165-168, 300-327, 350-362, 812-813, 845, 849-855, 886-891, 899
tests/functional_tests/run_func_test.py           527      2    99%   79, 177
-----------------------------------------------------------------------------
TOTAL                                            3271   1248    62%

Coverage report cannot identify the calls to jit-compiled functions.

XingerTang · 2026-07-08T20:18:07Z

The original accuracy tests have been split into two parts:

Minimalist accuracy test for CI/CD quality check
- Simplified report so we can focus on the most important values during production
- Much faster and lighter, so shorter delay for accuracy checks
- Still use the built distribution version for the OS-wise check
Accuracy benchmarking for monitoring the performance of combinations of different options
- Maximised report to cover as many details as possible
- Call the modules directly so that it runs faster and captures the debugging info/profiling info better
  - Run by from src.utils import run_full_accuracy_suite as run_full; run_full()

Still under development...

XingerTang · 2026-07-09T12:52:35Z

GitHub runner does not support Python 3.7 for ubuntu:

Error: The version '3.7' with architecture 'x64' was not found for Ubuntu 24.04.

for simplicity, just test version 3.9 to 3.13.
(AlphaImpute2 cross version testing is wrong, and requires a fix.)

…hr output

gregorgorjanc · 2026-07-31T18:17:43Z

+``-n_thread_fam`` controls the number of threads to parallelise accross
+families in a generation, while ``-n_thread_loci`` controls the number of threads 
+to parallelise across loci for each individual. Both options can be used together 
+to speed up the analysis, but ``-n_thread_fam`` is more efficient than 


Suggested change

to speed up the analysis, but ``-n_thread_fam`` is more efficient than

to speed up the analysis, but ``-n_thread_fam`` is more impactful than

gregorgorjanc · 2026-07-31T18:18:38Z

+to speed up the analysis, but ``-n_thread_fam`` is more efficient than 
+``-n_thread_loci`` when the number of loci is small (i.e., less than 3000
+per chromosome) and computing resources are limited.
+While the ``-n_thread_loci`` option might be more useful if number of CPU cores 


Suggested change

While the ``-n_thread_loci`` option might be more useful if number of CPU cores

The ``-n_thread_loci`` option might be more useful if number of CPU cores

gregorgorjanc

Minor suggested edits.

XingerTang · 2026-08-03T16:19:31Z

In ef647f0:

The section testing_and_profiling is moved from contribute to a separate page
- Its content has been updated according to the current code behaviour, and new details have been added
accuracy_report_comparison.py is added for the comparison of two accu_report

gregorgorjanc · 2026-08-03T16:28:41Z

The section testing_and_profiling is moved from contribute to a separate page

That's a good call @XingerTang! Just ensure you link it up well from contrib/dev page too.

gregorgorjanc

Nice, clear, and detailed instructions @XingerTang !

gregorgorjanc · 2026-08-03T16:32:43Z

The section testing_and_profiling is moved from contribute to a separate page

That's a good call @XingerTang! Just ensure you link it up well from contrib/dev page too.

I see that you have the linking, so all seems in order;)

XingerTang · 2026-08-04T18:15:39Z

The remaining things to do would be:

Pylint the other Python files
Update changelog

XingerTang · 2026-08-06T13:24:47Z

I'm not sure if we should be on the author list or the maintainer list. Could you review that?

gregorgorjanc · 2026-08-06T14:57:19Z

I'm not sure if we should be on the author list or the maintainer list. Could you review that?

I think your edits are in a good direction and make sense! Groovy;)

Please check with Ros if she would prefer Ros or Rosalind

Ensure you add space before Ros' surname;)

You misspelled alphagenes for Yu.

Best if you put me last in the author order.

Refactoring functional tests

eb8cdff

XingerTang marked this pull request as draft July 7, 2026 17:57

gregorgorjanc approved these changes Jul 7, 2026

View reviewed changes

Fix pytest warnings

0ef79a9

Simplify accuracy test report

5bd3c0c

gregorgorjanc approved these changes Jul 8, 2026

View reviewed changes

XingerTang added 3 commits July 9, 2026 13:26

Proper hybrid test

c187b62

Update test workflow

ced4322

Shorten the list of Python versions

4712661

Update flake8 configuration

eb1a6f2

gregorgorjanc approved these changes Jul 9, 2026

View reviewed changes

XingerTang added 6 commits July 9, 2026 16:10

Test older version Python with older OS images

583252b

Python 3.6 is not well-supported

b64d72b

Update requirement

3ea4652

Limit support to python >= 3.8

6360ffb

Fix rounding for alt_allele_prob output

58ff958

Ignore first two rows of seg files & special case for hap file of x_c…

badef09

…hr output

gregorgorjanc approved these changes Jul 9, 2026

View reviewed changes

XingerTang added 2 commits July 10, 2026 12:34

Add gen-wise abs_diff and correct_rate output of seg_prob

1f327ac

Refactor utils.py

6ded945

gregorgorjanc approved these changes Jul 10, 2026

View reviewed changes

Add SER and relevant metrics for hap

781f5c0

gregorgorjanc approved these changes Jul 10, 2026

View reviewed changes

Add runtime and visualization

9ed5406

gregorgorjanc reviewed Jul 31, 2026

View reviewed changes

Update and add new sections for testing_and_profiling docs

ef647f0

gregorgorjanc approved these changes Aug 3, 2026

View reviewed changes

Add profiling instruction & multi-thread test

6fd9054

gregorgorjanc approved these changes Aug 4, 2026

View reviewed changes

XingerTang added 2 commits August 4, 2026 19:03

Pylint test code

ab59965

Fix Github runner warnings

421e44f

gregorgorjanc approved these changes Aug 4, 2026

View reviewed changes

XingerTang added 4 commits August 5, 2026 12:46

Fix GitHub runner warning 2

5d5a083

Pylint conftest.py

54e983e

Pylint all files

334ee4a

Update Changelog

35fe751

XingerTang marked this pull request as ready for review August 5, 2026 14:39

gregorgorjanc approved these changes Aug 5, 2026

View reviewed changes

Bumped version to 1.4.0

ba3ff82

gregorgorjanc reviewed Aug 6, 2026

View reviewed changes

Comment thread pyproject.toml Outdated

gregorgorjanc approved these changes Aug 6, 2026

View reviewed changes

Update pyproject.toml and changelog

e474267

XingerTang force-pushed the refactor/refactor-AlphaPeel branch from 4594857 to e474267 Compare August 7, 2026 11:52

XingerTang merged commit a539a54 into AlphaGenes:devel Aug 7, 2026
18 checks passed

XingerTang deleted the refactor/refactor-AlphaPeel branch August 7, 2026 14:08

	to speed up the analysis, but ``-n_thread_fam`` is more efficient than
	to speed up the analysis, but ``-n_thread_fam`` is more impactful than

	While the ``-n_thread_loci`` option might be more useful if number of CPU cores
	The ``-n_thread_loci`` option might be more useful if number of CPU cores

Conversation

XingerTang commented Jul 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Related Issue

What changed

Bigger picture of the full plan

Why this change

Submodule changes

Notes / Risks

Uh oh!

XingerTang commented Jul 7, 2026

Uh oh!

gregorgorjanc left a comment

Choose a reason for hiding this comment

Uh oh!

XingerTang commented Jul 8, 2026

Uh oh!

XingerTang commented Jul 8, 2026

Uh oh!

XingerTang commented Jul 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

XingerTang commented Jul 9, 2026

Uh oh!

gregorgorjanc Jul 31, 2026

Choose a reason for hiding this comment

Uh oh!

gregorgorjanc Jul 31, 2026

Choose a reason for hiding this comment

Uh oh!

gregorgorjanc left a comment

Choose a reason for hiding this comment

Uh oh!

XingerTang commented Aug 3, 2026

Uh oh!

gregorgorjanc commented Aug 3, 2026

Uh oh!

gregorgorjanc left a comment

Choose a reason for hiding this comment

Uh oh!

gregorgorjanc commented Aug 3, 2026

Uh oh!

XingerTang commented Aug 4, 2026

Uh oh!

Uh oh!

XingerTang commented Aug 6, 2026

Uh oh!

gregorgorjanc commented Aug 6, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

XingerTang commented Jul 7, 2026 •

edited

Loading

XingerTang commented Jul 8, 2026 •

edited

Loading