Alphafold solution crashes client 2027

Started by Bletchley Park

Bletchley Park Lv 1

20210809-ba838e8d9b-win_x86

While testing an idea regarding Alphafold, it returned an invalid solution.
solution shared with scientists 'BP_crashes_alpha01'

I wanted to find out how reliable AF is, so I created a basic shape in 2027 with the sequence for Plant Seed Protein, assuming it would yield a high confidence. It returned a prediction with 39.8% confidence and 61.4% similarity. The resulting solution crashes the client. I would place my bets on the foldit score function rather than AF predictions.

This is the error log:

core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 466-1, 466-2, 466-3, 467-1!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 466-2, 466-3, 467-1, 467-2!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 475-3, 476-1, 476-2, 476-3!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 466-1, 466-2, 466-3, 467-1!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 466-2, 466-3, 467-1, 467-2!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 466-3, 467-1, 467-2, 467-3!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 475-1, 475-2, 475-3, 476-1!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 475-2, 475-3, 476-1, 476-2!
core.kinematics.AtomTree: [ ERROR ] No proper DoF can be found for these four atoms: 475-3, 476-1, 476-2, 476-3!

UNHANDLED EXCEPTION
1: RtlInitUnicodeString +499 bytes (no line)
2: RtlFreeHeap +126 bytes (no line)
3: HeapFree +20 bytes (no line)
4: _free_base +28 bytes (no line)
5: operator delete +11 bytes (no line)
6: boost::detail::sp_counted_impl_p::`scalar deleting destructor' +26 bytes (no line) 7: boost::detail::sp_counted_base::destroy +10 bytes (no line) 8: std::_Destroy_range1<std::allocator<boost::shared_ptr >,boost::shared_ptr *> +97 bytes (no line) 9: core::scoring::ResPairMinimizationData::~ResPairMinimizationData +81 bytes (no line) 10: core::scoring::MinimizationEdge::~MinimizationEdge +533 bytes (no line) 11: core::scoring::MinimizationEdge::`scalar deleting destructor' +11 bytes (no line) 12: core::scoring::MinimizationGraph::delete_edge +22 bytes (no line) 13: utility::graph::Graph::delete_everything +33 bytes (no line) 14: core::scoring::MinimizationGraph::~MinimizationGraph +56 bytes (no line) 15: core::scoring::MinimizationGraph::`scalar deleting destructor' +11 bytes (no line) 16: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 17: core::scoring::Energies::~Energies +407 bytes (no line) 18: core::scoring::Energies::`scalar deleting destructor' +11 bytes (no line) 19: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 20: core::pose::Pose::~Pose +507 bytes (no line) 21: core::pose::Pose::`scalar deleting destructor' +11 bytes (no line) 22: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 23: interactive::application::actions::cart::PoseLoopThreadActionCart::run_minimizer_on_work_pose +1263 bytes (no line) 24: interactive::application::actions::cart::ActionCartGlobalWiggle::thread_body_loop_unlocked +162 bytes (no line) 25: interactive::application::actions::cart::PoseLoopThreadActionCart::on_thread_run +4809 bytes (no line) 26: boost::function0::operator() +120 bytes (no line) 27: interactive::util::Thread::thread_entry_func +69 bytes (no line) 28: interactive::util::Thread::win32_thread_startup_helper +14 bytes (no line) 29: BaseThreadInitThunk +18 bytes (no line) 30: RtlInitializeExceptionChain +99 bytes (no line) 31: RtlInitializeExceptionChain +54 bytes (no line) Sending SOPs:

bkoep Staff Lv 1

Thanks for the crash report, Bletchley Park! I am able to reproduce this crash, so we should be able to track down the bug. However, from your error log, it looks like the client crashed in the middle of a Wiggle? Can you clarify, were you able to load the AlphaFold prediction without Wiggling, or did the client crash as soon as you tried to load the AlphaFold prediction?

Also, you bring up an important point about AlphaFold and natural proteins. In Foldit we are using an "abbreviated" version of AlphaFold that is not expected to work well on natural protein sequences.

The official, complete AlphaFold pipeline requires an extra step, scanning a large database for sequences that are similar to your query sequence. These similar sequences should all be evolutionarily related, and AlphaFold is able to extract patterns from this evolutionary data. AlphaFold is extremely good at extracting patterns from this evolutionary data, and this seems to be one of the reasons it performed so well in CASP.

When we use AlphaFold to predict Foldit designs, we skip this extra step because it is slow and because we do not expect to find "evolutionarily related" sequences for our designed proteins. Our internal benchmarking shows that AlphaFold is still good at predicting Foldit designed proteins, even though they don't have evolutionary data. However, skipping this step means that AlphaFold may underperform for natural protein sequences. If we were to provide evolutionary data for the plant seed protein, we could expect AlphaFold to produce a much more confident prediction.

Bletchley Park Lv 1

I was unable to load the AF result to begin with, it crashed while loading.
I was not wiggling in this thread. I just doubleclicked on the AF result in order to load it.
I have halted all other clients in other tracks and repeated the operation, this is the resulting exception:

UNHANDLED EXCEPTION
1: RtlInitUnicodeString +499 bytes (no line)
2: RtlFreeHeap +126 bytes (no line)
3: HeapFree +20 bytes (no line)
4: _free_base +28 bytes (no line)
5: operator delete +11 bytes (no line)
6: boost::detail::sp_counted_impl_p::`scalar deleting destructor' +26 bytes (no line) 7: boost::detail::sp_counted_base::destroy +10 bytes (no line) 8: std::_Destroy_range1<std::allocator<boost::shared_ptr >,boost::shared_ptr *> +97 bytes (no line) 9: core::scoring::ResPairMinimizationData::~ResPairMinimizationData +81 bytes (no line) 10: core::scoring::MinimizationEdge::~MinimizationEdge +533 bytes (no line) 11: core::scoring::MinimizationEdge::`scalar deleting destructor' +11 bytes (no line) 12: core::scoring::MinimizationGraph::delete_edge +22 bytes (no line) 13: utility::graph::Graph::delete_everything +33 bytes (no line) 14: core::scoring::MinimizationGraph::~MinimizationGraph +56 bytes (no line) 15: core::scoring::MinimizationGraph::`scalar deleting destructor' +11 bytes (no line) 16: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 17: core::scoring::Energies::~Energies +407 bytes (no line) 18: core::scoring::Energies::`scalar deleting destructor' +11 bytes (no line) 19: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 20: core::pose::Pose::~Pose +507 bytes (no line) 21: core::pose::Pose::`scalar deleting destructor' +11 bytes (no line) 22: boost::detail::sp_counted_impl_p::dispose +13 bytes (no line) 23: interactive::application::actions::cart::PoseLoopThreadActionCart::run_minimizer_on_work_pose +1263 bytes (no line) 24: interactive::application::actions::cart::ActionCartGlobalWiggle::thread_body_loop_unlocked +162 bytes (no line) 25: interactive::application::actions::cart::PoseLoopThreadActionCart::on_thread_run +4809 bytes (no line) 26: boost::function0::operator() +120 bytes (no line) 27: interactive::util::Thread::thread_entry_func +69 bytes (no line) 28: interactive::util::Thread::win32_thread_startup_helper +14 bytes (no line) 29: BaseThreadInitThunk +18 bytes (no line) 30: RtlInitializeExceptionChain +99 bytes (no line) 31: RtlInitializeExceptionChain +54 bytes (no line) Sending SOPs:

Nicm25 Lv 1

Yes, I also found solution that crashes when trying to load.
AF confidence is low so already I given up, but believe it will help and I will post.

I shared that solution 'SB_DFAFM002' to scientist in the [DEVPREV] AlphaFold Design puzzle.
AF After, they crashed when I tried to load them, and I could not load that results.
and puzzle 2027 also crashed when I had AF predict proteins with same AA's sequence.
(shared that solution '2027_DAF001' to scientist in 2027 puzzles)

use version :20210913-8a73606462-linux_x64-devprev