2020 年中值得關注的 AI 趨勢

https://chinese.classroom-aid.com/2020/06/2020-emerging-ai-trends.html/?fbclid=IwAR1H_Qj0yRIXg9GvK_jkdFciOneSD-4Dxg6eBC7iIFx5e8Faz24tthV-Eas

fatal: Could not read from remote repository. Please make sure you have the correct access rights and the repository exists.

https://segmentfault.com/a/1190000012141565

完勝ReLU！斯坦福的神經網絡採用這種激活函數，竟高保真還原各種圖像視頻

https://bangqu.com/vH1Xh6.html?fbclid=IwAR0VlzY39CLGZzn-NFCWWgxx6C8z-24IuFac_1IUcSEFKNnTUQQzkBKAfpU

Bert: Step by step by Hugging face

https://medium.com/swlh/bert-step-by-step-b7ff47fcfbe

Jupyter Notebook Cheat Sheet: A Beginner’s Guide to Jupyter Notebook

https://medium.com/edureka/jupyter-notebook-cheat-sheet-88f60d1aca7

Basic Geometry Optimization and Packmol Packaging Tutorial

https://www.youtube.com/watch?v=Bk96DOEhECw&feature=youtu.be&fbclid=IwAR1PNoK6cZ7ytL31_kNdcxeqfJonIUjkzg54fO42173iAwANF3l9_NBbXoU

MDAnalysis

https://www.mdanalysis.org/UserGuide/index.html

NAMD一些問題

1.能量最小化平衡的時候，如果沒有控制好NVT，會出現盒子變小，蛋白跑出水盒子。但是如果僅僅是控制體積不變，又會造成盒子內溶劑密度太小，產生‘Periodic cell has become too small for original patch grid！’這個錯誤。

NVT:

# Pressure and volume control

useGroupPressure       yes;         

useFlexibleCell        no;            

useConstantRatio       no;            

langevinPiston        on

langevinPistonTarget  1.01325 ;#  in bar -> 1 atm

langevinPistonPeriod  200.

langevinPistonDecay   50.

langevinPistonTemp    $temperature

在

wrapWater           on

wrapAll             on

這後邊加一個

margin 3

we use the margin 3 command to allow large volume fluctuations that are typical during the first dynamics of a new system in an NpT ensemble.

Fatal error on PE 0> FATAL ERROR: UNABLE TO FIND ANGLE PARAMETERS FOR HT

HT OT (ATOMS 5525 5527 5526)

錯誤原因：水分子的兩個HT原子之間不應該成鍵

TIP3P模型中，shake算法的水分子HH應該是有鍵的，但是vmd中那個solvate模塊自動加的水的HH是沒有鍵的，主動改一下solvate的top文件

修改辦法：將原文件中的結晶水刪除

原因：原體系中存在結晶水

1)

關於FATAL ERROR: Periodic cell has become too small for original patch grid!

幾個解決方法:

a.        將實驗環境改成在 NVT 或 NVE 條件下先進行平衡一段時間，之後視需要再改成 NPT 就較不容易出錯。(主要方法!)

b.        langevinPistonPeriod  至少400，langevinPistonDecay至少200，有必要的話加到1000 500 也行

c.        增加 margin

 (2)

關於 NPT NVT NVE 的 conf文件設置:

NPT:在 conf 文件添加關於 langevinpiston (定壓)和 langevin (定溫)兩項參數以及衍生參數的設置

NVT:取消 langevinpiston (定壓) 等相關參數的設置(設爲off或在最前面加#符號都行)，langevin (定溫) 參數維持不變。

NVE:取消 langevinpiston (定壓)和 langevin (定溫) 等相關參數的設置

蛋白質跑到水盒子外

cellBasisVector是按照水的minmax寫的

Set water [atomselect top water]

Measure minmax $water

Cellprorigin 測量top all 的center

PME比cell稍微大

出現的原因是蛋白質距離膜邊界太近

ERROR: Constraint failure in RATTLE algorithm for atom 21750!

ERROR: Constraint failure; simulation has become unstable.

出現原因邊界條件選擇不合適
 

在生成psf文件時要考慮二硫鍵的存在，可使用自動生成工具；

Keep-water-out可使用默認的參數；

小分子的參數文件要添加到蛋白質-膜的參數文件中，在線生成的小分子psf文件要轉換成charmm力場格式的。

參考

https://www.twblogs.net/a/5c2672a8bd9eee16b4a81683

VMDNAMD命令規則

affect 如何判斷 tip ant ready 命令模式 dba pac The

輸出體系的整個帶電量：
measure sumweights $all weight charge

給PDB文件設置周期邊界條件：
pbc set {54 54 24 } -all

將此晶胞內原子脫除周期性
pbc unwrap

顯示晶胞邊界
pbc box

體系總質量：
measure sumweights $all weight mass

體系總電荷：
measure sumweights $all weight charge

添加周期性水環境
package require solvate
solvate ubq.psf ubq.pdb -t 5 -o ubq wb
-t (override with any of the following)
-o (data will be written to output.psf/output.pdb)
添加離子中和多余電荷
autoionize -psf file.psf -pdb file.pdb [options]
可以直接使用VMD中extension>modeling>add ions的autoionize完成
TK console中選擇全部原子，保存psf與pdb文件
參考：Autoionize Plugin
命令模式運行ubq.pgn
> vmd -dispdev text -e ubq.pgn
測量周期最大最小邊界
set everyone [atomselect top all]
measure minmax $everyone
測量周期中心點
set everyone [atomselect top all]
measure center $everyone
測量質量中心
measure center $sel weight mass
位置移動
atomselect0 moveby {1 1 6} # 把所選原子向1，1，6向量方向和距離上移動
atomselect0 moveto { 3 6 5} # 把所選內容移動到3，6，5位置
原文：VMD的TK Console中的內置命令
設置resname/chain/resid
mol load pdb fileA.pdb
set sel [atomselect top "serial <=9"]
$sel set resname CD
$sel set chain X
$sel set resid 8
set all [atomselect top all]
$all writepdb fileB.pdb
refer the VMD Graphical Representations>Selections>Keyword for more values
top文件
IC A B C D [bond(AB)] [angle(ABC)] [dihedral(ABCD)] [angle(BCD)] [bond(CD)]
IC A B *C D [bond(AC)] [angle(BCA)] [improper(ABCD)] [angle(BCD)] [bond(CD)].
The * next to the C atom indicates that it is at the center of an improper angle definition. No * indicates that the 4 atoms do not have an improper topology.
Specifying IC’s is not necessary if you already have all the atoms in your PDB file. The entries are simply there
to have a way to create missing atoms from the positions of present ones.
參考：NAMD topology-tutorial
VMD鍵長單位
number in the “Value” field corresponds to the length of the bond in ?ngstroms.
reference:Using VMD - An Introductory Tutorial
psfgen命令
pdbalias residue HOH TIP3 #aliasing residue HOH to TIP3
pdbalias atom TIP3 O OH2 #aliasing residue TIP3 atom O to OH2
vmd中tk命令打開文件
mol new filename.psf #打開一個文件
mol addfile filename.pdb #打開另一個文件，並疊加到第一個文件結構上
最適分子數
Presently, the maximum number of atoms one can expect to realistically compute with high level quantum chemistry is approximately 120 atoms
平衡態模擬
常見的模擬思路是，先在NVT下約束住你的溶質（劑）做限制性模擬，這是一個升溫的過程，當溫度達到你的設定後, 接著做NPT模擬，此過程將調整體系的壓強進而使體系密度收斂。
經過一段時間的平衡模擬，在確定系統弛豫已經完全消除之後，就可以開始取數據了。如何判斷體系達到平衡，簡單的講可以通過以 下幾種方式，一，看能量（勢能，動能和總能）是否收斂；二，看系統的壓強，密度等等是否收斂；三看系統的RMSD是否達到你能接受的範圍，等等。
升溫/降溫控制
# IF Heating
reassignFreq 2000
reassignTemp 100
reassignIncr 1
reassignHold 300
#從100K升溫到300K，每2000步升高1K

# IF Cooling
reassignFreq 2000
reassignTemp 300
reassignIncr -1
reassignHold 100
固定分子
all atoms with a value of 1 (or a number different of 0) in a predetermined column will be fixed; atoms with a value of 0 in the same column will not be affected.
load pdb file into the psf file in vmd
in TKcon
set allatoms [atomselect top all]
$allatoms set beta 0
set fixedatom [atomselect top "resid 1 and name CA"]
$fixedatom set beta 1
$allatoms writepdb fileName.ref
拉動原子
to set which atom is to be pulled (SMD atom).uses the occupancy column of the pdb file to distinguish it
set allatoms [atomselect top all]
$allatoms set occupancy 0
set smdatom [atomselect top "resid 76 and name CA"]
$smdatom set occupancy 1
$allatoms writepdb fileName.ref
牽拉方向
set smdpos [lindex [$smdatom get {x y z}] 0]
set fixedpos [lindex [$fixedatom get {x y z}] 0]
vecnorm [vecsub $smdpos $fixedpos]

VMD計算SASA(solvent-accessible surface area )

set all [atomselect top "all"]
set some [atomselect top "resid 1 to 5"]
measure sasa 1.4 $all -restrict $some -points sasapoints
foreach pt $sasapoints {
draw point $pt
}
以上可以在窗口中顯示出1-5號氨基酸的溶劑可及表面，以point繪出。當然如果不需要瀏覽僅僅計算的話，只需要：
set all [atomselect top "all"]
set some [atomselect top "resid 1 to 5"]
measure sasa 1.4 $all -restrict $some
如果想看一下準不準可以算一下除了1-5號氨基酸的是多少
set someothers [atomselect top "not(resid 1 to 5)"]
measure sasa 1.4 $all -restrict $someothers
將兩個值相加，看是否是
measure sasa 1.4 $all
得到的結果。

原文摘錄自：關於溶劑可及表面積（Solvent Accessible Surface）- Biocheming
try VMD timeline Tool to calculate the change in SASA over time.
Load the molecule and trajectory and then go to:
Extensions -> Analysis -> Timeline.
SASA is one of the many parameters you can calculate in the Timeline window.
2014-9-10
原文摘錄自：How to calculate solvent accessible surface (SASA)?-Uttam Pal
[筆記]VMD/NAMD命令/規則
此博文包含圖片 (2013-10-10 16:18:42)
轉載▼
標簽：
vmd
namd
筆記
分類： 分子模擬
添加周期性水環境
package require solvate
solvate ubq.psf ubq.pdb -t 5 -o ubq wb
-t (override with any of the following)
-o (data will be written to output.psf/output.pdb)
添加離子中和多余電荷
autoionize -psf file.psf -pdb file.pdb [options]
可以直接使用VMD中extension>modeling>add ions的autoionize完成
TK console中選擇全部原子，保存psf與pdb文件
參考：Autoionize Plugin
命令模式運行ubq.pgn
> vmd -dispdev text -e ubq.pgn
測量周期最大最小邊界
set everyone [atomselect top all]
measure minmax $everyone
測量周期中心點
set everyone [atomselect top all]
measure center $everyone
測量質量中心
measure center $sel weight mass
位置移動
atomselect0 moveby {1 1 6} # 把所選原子向1，1，6向量方向和距離上移動
atomselect0 moveto { 3 6 5} # 把所選內容移動到3，6，5位置
原文：VMD的TK Console中的內置命令
設置resname/chain/resid
mol load pdb fileA.pdb
set sel [atomselect top "serial <=9"]
$sel set resname CD
$sel set chain X
$sel set resid 8
set all [atomselect top all]
$all writepdb fileB.pdb
refer the VMD Graphical Representations>Selections>Keyword for more values
top文件
IC A B C D [bond(AB)] [angle(ABC)] [dihedral(ABCD)] [angle(BCD)] [bond(CD)]
IC A B *C D [bond(AC)] [angle(BCA)] [improper(ABCD)] [angle(BCD)] [bond(CD)].
The * next to the C atom indicates that it is at the center of an improper angle definition. No * indicates that the 4 atoms do not have an improper topology.
Specifying IC’s is not necessary if you already have all the atoms in your PDB file. The entries are simply there
to have a way to create missing atoms from the positions of present ones.
參考：NAMD topology-tutorial
VMD鍵長單位
number in the “Value” field corresponds to the length of the bond in ?ngstroms.
reference:Using VMD - An Introductory Tutorial
psfgen命令
pdbalias residue HOH TIP3 #aliasing residue HOH to TIP3
pdbalias atom TIP3 O OH2 #aliasing residue TIP3 atom O to OH2
vmd中tk命令打開文件
mol new filename.psf #打開一個文件
mol addfile filename.pdb #打開另一個文件，並疊加到第一個文件結構上
最適分子數
Presently, the maximum number of atoms one can expect to realistically compute with high level quantum chemistry is approximately 120 atoms
平衡態模擬
常見的模擬思路是，先在NVT下約束住你的溶質（劑）做限制性模擬，這是一個升溫的過程，當溫度達到你的設定後, 接著做NPT模擬，此過程將調整體系的壓強進而使體系密度收斂。
經過一段時間的平衡模擬，在確定系統弛豫已經完全消除之後，就可以開始取數據了。如何判斷體系達到平衡，簡單的講可以通過以 下幾種方式，一，看能量（勢能，動能和總能）是否收斂；二，看系統的壓強，密度等等是否收斂；三看系統的RMSD是否達到你能接受的範圍，等等。
升溫/降溫控制
# IF Heating
reassignFreq 2000
reassignTemp 100
reassignIncr 1
reassignHold 300
#從100K升溫到300K，每2000步升高1K

# IF Cooling
reassignFreq 2000
reassignTemp 300
reassignIncr -1
reassignHold 100
固定分子
all atoms with a value of 1 (or a number different of 0) in a predetermined column will be fixed; atoms with a value of 0 in the same column will not be affected.
load pdb file into the psf file in vmd
in TKcon
set allatoms [atomselect top all]
$allatoms set beta 0
set fixedatom [atomselect top "resid 1 and name CA"]
$fixedatom set beta 1
$allatoms writepdb fileName.ref
拉動原子
to set which atom is to be pulled (SMD atom).uses the occupancy column of the pdb file to distinguish it
set allatoms [atomselect top all]
$allatoms set occupancy 0
set smdatom [atomselect top "resid 76 and name CA"]
$smdatom set occupancy 1
$allatoms writepdb fileName.ref
牽拉方向
set smdpos [lindex [$smdatom get {x y z}] 0]
set fixedpos [lindex [$fixedatom get {x y z}] 0]
vecnorm [vecsub $smdpos $fixedpos]
run a NAMD simulation
namd2 +p[procs] configfile > outfile
力場中二面角倒序正序區別
軟件裏的二面角有正負是得考慮4號原子在123號原子平面的哪一側，舉MS裏二面角的來說，按順序點1234共4個原子，二面角指的是123平面和234 平面的二面角，沿著23的軸看過去，1號原子在4號原子的逆時針方向（左手邊），則取正值，反之取負值；絕對值仍在[0,180]。
原文摘錄自：二面角和扭轉角區別
Reversing the atoms in a proper dihedral does not have any mathematical or practical consequences. Conversely, the ordering of the atoms is very important for improper dihedrals.
原文摘錄自：Parameters for a Protein-RNA covalent bond

力場參數圖示
[筆記]VMD/NAMD命令/規則2014-06-18
圖片來源：Brève introduction à la mécanique (MM) et à la dynamique moléculaire (DM)

improper
dihedral

By convention, the first atom of an improper
dihedral (type A-X-X-B or A-B-C-D) is usually the central atom. This had
been a general rule in the past.

2014-06-19

原文摘錄自：CHARMM c32b2 parmfile.doc
拓撲文件AUTOGENERATE ANGLES DIHEDRAL
AUTOgenerate default options to be used when building a sturcture.AUTO ANGLes specifies that all possible angles and DIHEdral specifies that all possible dihedral angles be generated when building a structure. If these options are not included the angles and/or dihedrals must be listed explicitly in the topology file
2014-06-19
原文轉自：CHARMM c38b1 rtop.doc

namd開始NPT計算過程報錯Periodic cell has become too small for original patch grid解決方法
I have a different approach to fixing this problem. I believe that the underlying cause of this error is that the system is resizing too quickly (due to the system pressure being far from the specified pressure). To slow the resizing down I increase langevinPistonPeriod to 1000 and langevinPistonDecay to 500 for a short time when I start the NPT simulation, and then switch to the normal values of 100 and 50.

For example, in your case I would do one input file for 100 ps with a constraintscaling of 10, langevinPistonPeriod of 1000, and a langevinPistonDecay of 500. Then I would load the restart files into a new simulation and proceed as you were above. Also, by restarting the simulation you are forcing NAMD to re-assign the patches, which will help prevent this error.
2014-08-26
原文摘錄自：Re: Periodic cell has become too small for original patch grid!
FATAL ERROR: Periodic cell has become too small for original patch grid!
問題出現在nvt計算完成後第一次進行npt的時候.通過在namd的郵件列表尋找，最後發現是周期性盒子設置的時候四舍五入數值，造成盒子在npt計算 過程中溶劑環境密度過小（盒子體積過大，而分子總數一定，最終水溶液密度過低）。解決方法，周期盒子大小不能直接四舍五入，要考慮到盒子體積與分子總數得 到溶液的密度是否恰當好處。改用該方法後，順利計算，為提示錯誤。
2014-9-10
該內容原創

計算namd體系密度
首先計算體系的質量，然後根據namd輸出的體積，兩者相除即可。過程如下：
例如：
C 11
H 1596
O 790
N 2
上面表示的是每種原子的個數
則mass=（11*12+1596*1+790*16+2*14）/NA 單位為克。
NA為阿伏伽德羅常數：6.02214129(27)×1023
那麽密度為mass/V
2014-08-26
原文摘錄自：heating過程中水盒子形狀發生變化

VMD計算SASA(solvent-accessible surface area )
控制臺下用命令即可得出：
set all [atomselect top "all"]
set some [atomselect top "resid 1 to 5"]
measure sasa 1.4 $all -restrict $some -points sasapoints
foreach pt $sasapoints {
draw point $pt
}
以上可以在窗口中顯示出1-5號氨基酸的溶劑可及表面，以point繪出。當然如果不需要瀏覽僅僅計算的話，只需要：
set all [atomselect top "all"]
set some [atomselect top "resid 1 to 5"]
measure sasa 1.4 $all -restrict $some
如果想看一下準不準可以算一下除了1-5號氨基酸的是多少
set someothers [atomselect top "not(resid 1 to 5)"]
measure sasa 1.4 $all -restrict $someothers
將兩個值相加，看是否是
measure sasa 1.4 $all
得到的結果。
2014-9-10
原文摘錄自：關於溶劑可及表面積（Solvent Accessible Surface）- Biocheming
try VMD timeline Tool to calculate the change in SASA over time.
Load the molecule and trajectory and then go to:
Extensions -> Analysis -> Timeline.
SASA is one of the many parameters you can calculate in the Timeline window.