MAINMASTseg

MAINMAST is a de novo modeling protocol to build an entire protein 3D model directly from near-atomic resolution EM map.
MAINMAST is a fully automated protocol and can generate reliable initial C-alpha models which can be used to construct full atomic models. This new de novo modeling method has several advantages; (1) It does not require reference structures; (2) It does not requre manual interventions; (3) a pool of candidate models are generated.

Introduction

MAINMASTseg protocol

MAINMAST protocol consists of mainly four steps:

(1) Identify local dense points (LDPs) by Mean Shifting Algorithm;

(2) Detect symmetrically equivarent positions in LDPs;

(3) Connect all LDPs by Minimum Spanning Trees (MSTs);

(4) Traceback segmented MSTs to the EM map;

Tutorial

Protocol

MAINMASTseg is an Automated segmentation program for EM maps of near atomic resolution (less than 4.5 angstrom) with symmetry

MAINMASTseg protocol consists of mainly four steps:
(1) Identify local dense points in an EM map by Mean Shifting clustering algorithm;
(2) Detect symmetrycally equivalent LDPs (tabu-pairs);
(3) Connect all LDPs by Minimum Spanning Tree;
(4) Traceback labeled LDPs and generate segmented density maps;

Program MAINMASTseg will perform all the (1)-(4) steps.

Flow Chart of MAINMAST

1. Identifying local dense points

Each Grid points in a EM map are clusterized by a non-parametric clustering algorithm (Mean shift). After the clustering, the representative points in the clusters are called local dense points (LDPs).

Map (EMD-0093) and LDPs

2. Detect symmetrycally equivalent LDPs (tabu-pairs)

MAINMASTseg identidies the correspondence in stmmetrically equivalent LDPs by considering rotation matrix. The pair of LDPs should not be included in the same chain.

3. Connect all LDPs by Minimum Spanning Tree

Minimum Spanning Tree is a graph structure that connects all vertices with the minimum total length of edges.

Building of MSTs
MSTs

4. Traceback labeled LDPs and generate segmented density maps.

The segmented MSTs are converted to segmented density maps.

Pgreenicted path

Commands

Commands

MAINMASTseg and some tools.

MAINMASTseg

Mainm program. MAINMASTseg computes segmented density maps (mrc format) and builds MST model in CIF format.

Usage: MainmastSeg -i [MAP.mrc] -Y [Rotation Matrix] [(option)] Fast MainmastSeg C-lang & multi thread version ---Mode--- -L : Fast LDP search mode -M : Minimum Spanning Tree Mode (default) -G : Graph Mode -W : Generating MRC file Mode. File name segment%d.mrc -V : Movie Mode ---Options--- -c [int ] :Number of cores for threads def=2 -t [float] :Threshold of density map def=0.000 -g [float] : bandwidth of the gaussian filter def=2.0, sigma = 0.5*[float] -f [float] :Filtering for representative points def=0.100 -m [float] :After MeanShifting merge LDPs where d<[float] def=0.500 -R [float] :Radius of Local MST def=10.000 -k [float] :keep edges where d < [float] def=0.500 Thi is Ver 1.041

bondtreeCIF.pl

This program makes a Pymol script for visualization of MSTs.

bondtreeCIF.pl [Output of MAINMASTseg (CIF format)] > a.txt

Then, the "a.txt" can be used by pymol as

pymol -u a.txt

conv_ncs.pl

This program convert symmetry_from_map.ncs_spec (output file of phenix.map_symmetry) to the rotation matrix file.

conv_ncs.pl [symmetry_from_map.ncs_spec] > MTX.txt

Examples

Example1: EMD-0093 (3.4 angstrom resolution)

(Optional) Remove noise from the EM map by UCSF Chimera

Some EM map contains many noise at the recommended contour level. Before computing segmentation, noise can be removed hideDust command, in UCSF Chimera.

sop hideDust #0 size 100 metric volume

Then save as "MAP_m4A.mrc".

MAP_m4A.mrc (generated from EMD-0093) at the density threshold=0.7

Make a rotation matrix file (MTX.txt) by UCSF Chimera

EMD-0093 has C4 symmetry. The size is (220*1.34,220*1.34,220*1.34). The center is (147.4,147.4,147.4).
1. open 6gyn.pdb as #0
2. Type the following command in Chimera command line

sym #0 group c4 center 147.4,147.4,147.4

3. Save all pdbs as symmetry.pdb 4. Extract the rotation matrix from the saved pdb file as:

grep BIOMT[1-3] symmetry.pdb > MTX.txt

(Optional) The MTX.txt should be the following format:

REMARK 350   BIOMT1   1  1.000000  0.000000  0.000000        0.00000            
REMARK 350   BIOMT2   1  0.000000  1.000000  0.000000        0.00000            
REMARK 350   BIOMT3   1  0.000000  0.000000  1.000000        0.00000            
REMARK 350   BIOMT1   2  0.000000 -1.000000  0.000000      294.80000            
REMARK 350   BIOMT2   2  1.000000  0.000000  0.000000        0.00000            
REMARK 350   BIOMT3   2 -0.000000  0.000000  1.000000        0.00000            
REMARK 350   BIOMT1   3 -1.000000 -0.000000  0.000000      294.80000            
REMARK 350   BIOMT2   3  0.000000 -1.000000  0.000000      294.80000            
REMARK 350   BIOMT3   3  0.000000  0.000000  1.000000        0.00000            
REMARK 350   BIOMT1   4 -0.000000  1.000000  0.000000        0.00000            
REMARK 350   BIOMT2   4 -1.000000 -0.000000  0.000000      294.80000            
REMARK 350   BIOMT3   4  0.000000  0.000000  1.000000        0.00000

Make a rotation matrix file (MTX.txt) by phenix.map_symmetry

When the center of the EM map is unknown. phenix.map_symmetry is able to identify the rotation matrix.

phenix.map_symmetry MAP_m4A.mrc symmetry=C4

Then, convert the output file (symmetry_from_map.ncs_spec) to MTX.txt.

../conv_ncs.pl symmetry_from_map.ncs_spec > MTX.txt

Segmentation by MAINMASTseg

MAINMASTseg generates the segmented MST (-M option) and density maps (-W option).
1. Generate MSTs only with the recommended contour level 0.7.

../MainmastSeg -i MAP_m4A.mrc -Y MTX.txt -c 8 -t 0.7 -M > test.cif

Visualize MSTs by Pymol.

../bondtreeCIF.pl test.cif > a.txt

Open a.txt by pymol:

pymol -u a.txt

LDPs

segmented MSTs

Generate segmented density maps

Once you confirmed the MSTs, MAINMASTseg can generate the segmented density maps with -W option:

../MainmastSeg -i MAP_m4A.mrc -Y MTX.txt -c 8 -t 0.7 -M -W > test2.cif

This command will generate 4 mrc-format files (region0.mrc,region1.mrc,region2.mrc,region3.mrc).

Segmented density maps

Download

MAINMASTseg programs:

  • MAINMASTseg is available at Github

Updates of MAINMASTseg programs:

  • 2019 11/21 Released Version 1.0




Tech Specs


CPU: >=4 cores
Memory: >=20Gb
GPU: not required.

FAQ

For MainMastseg FAQ click here


License

© 2019 Genki Terashi, Daisuke Kihara and Purdue University

MAINMASTseg is a free software for academic and non-commercial users.
It is released under the terms of the GNU General Public License Ver.3 (https://www.gnu.org/licenses/gpl-3.0.en.html).
Commercial users please contact dkihara@purdue.edu for alternate licensing.



Reference