0000363: Memory leak during ML phylogenetic analysis

View Issue Details [ Jump to Notes ]

[ Issue History ] [ Print ]

Project

Category

View Status

Date Submitted

Last Update

0000363

MEGA

Main Form

public

2017-01-13 11:34

2017-01-27 13:21

Reporter

guest

Assigned To

gstecher

Priority

normal

Severity

minor

Reproducibility

have not tried

Status

resolved

Resolution

no change required

Platform

Windows

Product Version

MEGA 11 (Graphical Interface version)

Target Version

Fixed in Version

Summary

0000363: Memory leak during ML phylogenetic analysis

Description

Windows 10 x64 (8 gb RAM) wants to shut MEGA down due to lack of memory about halfway through ML phylogenetic analysis. I increased pagefile to 32 gb and it still ran out of physical and pagefile virtual memory, completing only 109 of 250 bootstraps. Note: Closing the analysis windows such that MEGA GUI is all that is running still does not release memory! Once you exit MEGA then memory is released. This problem only occurs when selecting Construct/Test Maximum-likelihood tree. The other tree building methods do not cause the problem. Memory is also not released when aborting tree building.

The analysis uses 56 sequences with 1142 nucleotides.

Evolutionary model does not matter.

Steps To Reproduce

Problem only occurs during Construct/Test Maximum-likelihood tree. All models exhibit the problem. The only setting which appears to allow construction to finish is to mitigate the problem by reducing SPR from 5 to 3, and setting Branch Swap Filter to weak. However, even when the tree is finished, memory is still not released. If you Stop analysis, memory is not released.

Analysis
  Analysis ------------------------ Phylogeny Reconstruction
  Statistical Method -------------- Maximum Likelihood
Phylogeny Test
  Test of Phylogeny --------------- Bootstrap method
  No. of Bootstrap Replications --- 250
Substitution Model
  Substitutions Type -------------- Nucleotide
  Model/Method -------------------- Tamura-Nei model
Rates and Patterns
  Rates among Sites --------------- Gamma Distributed (G)
  No of Discrete Gamma Categories - 16
Data Subset to Use
  Gaps/Missing Data Treatment ----- Complete deletion
Tree Inference Options
  ML Heuristic Method ------------- Subtree-Pruning-Regrafting - Extensive (SPR level 5)
  Initial Tree for ML ------------- Make initial tree automatically (Default - NJ/BioNJ)
  Branch Swap Filter -------------- None
System Resource Usage
  Number of Threads --------------- 4
Codons Included ------------------- 1st+2nd+3rd+Non-Coding

Additional Information

I've attached the sequence data for you to explore. I've enjoyed using MEGA for several years. Hope the problem is resolved soon.

Tags

No tags attached.

Attach Tags

(Separate by ",")

First Name

Nick

Last Name

Waters

nickdwaters@gmail.com

Confirm Email

nickdwaters@gmail.com

Attached Files

cytb_cleaned_aligned_with_Nielsen_and_outgroups_trimmed reduced ASMO v4.mas (66,396 bytes) 2017-01-13 11:34

Relationships

Notes
(0003749) gstecher (administrator) 2017-01-27 13:21	Hi Nick, I am writing in response to the bug report you recently submitted regarding the MEGA software. To reduce the amount of memory used for your analysis, you should reduce the number of gamma categories used (you are using 16 where as the default is 5) as that setting increases the amount of RAM required linearly. -- Best regards, Glen Stecher Institute for Genomics and Evolutionary Medicine igem.temple.edu

Issue History
Date Modified	Username	Field	Change
2017-01-13 11:34	guest	New Issue
2017-01-13 11:34	guest	File Added: cytb_cleaned_aligned_with_Nielsen_and_outgroups_trimmed reduced ASMO v4.mas
2017-01-27 13:21	gstecher	Note Added: 0003749
2017-01-27 13:21	gstecher	Status	new => resolved
2017-01-27 13:21	gstecher	Resolution	open => no change required
2017-01-27 13:21	gstecher	Assigned To	=> gstecher