From: Miguel <mi...@jm...> - 2004-11-30 13:52:57
|
Some months ago there was discussion of 'color group' I am now ready to pick up that thread and work on it. RasMol calculates the colors by getting the min and max group numbers across all chains. Groups are then scaled according to that range. RasMol does not handle insertion codes. Q: Should insertion codes receive any special treatment when calculating group colors? Miguel |
From: timothy d. <mol...@ma...> - 2004-11-30 14:35:51
|
On 2004-11-30 (08:52) Miguel wrote: >Some months ago there was discussion of 'color group' > >I am now ready to pick up that thread and work on it. > hi Miguel, cool - that would be great. >RasMol calculates the colors by getting the min and max group >numbers across all chains. Groups are then scaled according to that >range. > I wrote a Web page that calculates various gradient styles: <http://www.molvisions.com/resources/color_gradients/> some issues I encountered along the way: 1. in Rasmol, one can not apply 'color group' on a single chain of a multi-= chain structure. It would be nice if Jmol applied the color scheme to the = selected set only. this would provide more flexibility in how it can be us= ed. 2. a choice of gradient styles would be nice. Rasmol uses a 'rainbow' of = sorts, which can be useful in the right context. other suggestions: blue-w= hite-red (N to C), and grayscale. 3. too many colors in the gradient, and RasMol (and Chime) runs out of colo= rs (can not allocate shade!). I don't think this will bother Jmol, though. >RasMol does not handle insertion codes. > >Q: Should insertion codes receive any special treatment when >calculating group colors? > I can't see any reason. but...I am not one to ask. ;-) regards, tim --=20 Timothy Driscoll molvisions - see, grasp, learn. <http://www.molvisions.com/> usa:north carolina:wake forest "I sometimes go to my own little world, but that's okay, they know me there= =2E" |
From: Miguel <mi...@jm...> - 2004-11-30 15:12:49
|
Tim, I just checked in code which is RasMol compatible on 'color group'. v10pr= e20b > 1. in Rasmol, one can not apply 'color group' on a single chain of a > multi-chain structure. It would be nice if Jmol applied the color sche= me > to the selected set only. this would provide more flexibility in how i= t > can be used. Yes, it seems that would be much more useful. Q: Is incompatibility going to be a problem? > 2. a choice of gradient styles would be nice. Rasmol uses a 'rainbow'= of > sorts, which can be useful in the right context. other suggestions: > blue-white-red (N to C), and grayscale. We would need to introduce other keywords for this. color groupBlueWhiteRed color groupGrayscale The alternative would be a 'set' command. > 3. too many colors in the gradient, and RasMol (and Chime) runs out of > colors (can not allocate shade=21). I don't think this will bother Jmo= l, > though. Not a problem for Jmol. Miguel |
From: timothy d. <mol...@ma...> - 2004-11-30 15:34:23
|
On 2004-11-30 (10:12) Miguel wrote: >Tim, > >I just checked in code which is RasMol compatible on 'color group'. >v10pre20b > cool! >>1. in Rasmol, one can not apply 'color group' on a single chain of a >>multi-chain structure. It would be nice if Jmol applied the color >>scheme to the selected set only. this would provide more flexibility >>in how it can be used. > >Yes, it seems that would be much more useful. > >Q: Is incompatibility going to be a problem? > yeah, it is. :-( can we implement a Jmol-specific variety of the command? like 'color gradi= ent'? >>2. a choice of gradient styles would be nice. Rasmol uses a=20 >'rainbow' of >>sorts, which can be useful in the right context. other >>suggestions: blue-white-red (N to C), and grayscale. > >We would need to introduce other keywords for this. > >color groupBlueWhiteRed color groupGrayscale > >The alternative would be a 'set' command. > a 'set' command would be cleaner IMHO. which do you prefer? tim --=20 Timothy Driscoll molvisions - see, grasp, learn. <http://www.molvisions.com/> usa:north carolina:wake forest "Enzymes are things invented by biologists that explain things=20 which otherwise require harder thinking." |
From: Eric M. <em...@mi...> - 2004-11-30 17:09:38
|
At 11/30/04, you wrote: >can we implement a Jmol-specific variety of the command? like 'color >gradient'? I agree this is a simple solution that leaves 'color group' RasMol/Chime compatible. >a 'set' command would be cleaner IMHO. which do you prefer? Why not 'color gradient' (with default to rainbow) or 'color gradient rainbow' 'color gradient grayscale' avoiding yet another set command? /* - - - - - - - - - - - - - - - - - - - - - - - - - - - Protein Explorer - 3D Visualization: http://proteinexplorer.org Workshops: http://www.umass.edu/molvis/workshop Biochem Structure Tutorials http://MolviZ.org World Index of Molecular Visualization Resources: http://molvisindex.org ConSurf - Find Conserved Patches in Proteins: http://consurf.tau.ac.il Atlas of Macromolecules: http://molvis.sdsc.edu/atlas/atlas.htm PDB Lite Macromolecule Finder: http://pdblite.org Molecular Visualization EMail List (molvis-list): http://bioinformatics.org/mailman/listinfo/molvis-list Eric Martz, Professor Emeritus, Dept Microbiology U Mass, Amherst MA US 413-545-2325/FAX 413-545-2532 http://www.umass.edu/molvis/martz - - - - - - - - - - - - - - - - - - - - - - - - - - - */ |
From: Miguel <mi...@jm...> - 2004-11-30 18:50:00
|
> At 11/30/04, you wrote: >>can we implement a Jmol-specific variety of the command? like 'color >>gradient'? > > I agree this is a simple solution that leaves 'color group' RasMol/Chim= e > compatible. There are a couple of issues that need to be separated. Issue 1 is the question of what the range is for the scale. There are three cases that I can think of. 1. absolute - formalCharge - partialCharge - fixedTemperature 2. relative to all the atoms in the model - relativeTemperature - group 3. relative to the selected set We don't have anything that currently maps the scale to the currently selected set ... that is the first problem that we are trying to solve. Issue 2 is what the color gradient is. This is independent of which numeric metric you are trying to map. I would like like to reserve the word 'gradient' for a general-purpose gradient that could be applied to anything ... not just group sequence number. What I am thinking is that the 'gradient' command defines a color gradien= t that is subsequently used by other 'color =7Bcolor-scale=7D' commands. >>a 'set' command would be cleaner IMHO. which do you prefer? > > Why not 'color gradient' (with default to rainbow) or > 'color gradient rainbow' > 'color gradient grayscale' > avoiding yet another set command? If we can come up with a single word, I would rather not deal with a variable number of arguments. Leaving the 'gradient' issue aside for a minute ... Can we come up with a single word that describes this? sequenceNumber? groupNumber? residueNumber? Miguel |
From: Eric M. <em...@mi...> - 2004-11-30 17:20:43
|
At 11/30/04, you wrote: >Q: Should insertion codes receive any special treatment when calculating >group colors? The main uses of 'color group', in my experience, are (i) to facilitate tracing the path of a chain with the eye (ii) to make it easy to see the N and C termini (or 3' and 5' termini). I don't think special treatment for insertion-coded residues would be useful. For the rare case where it might, the user could write a script. ---------------- The fact that RasMol/Chime (R/C) use a single gradient, pegged to sequence numbers, for all chains, is unfortunate. Typically, one long chain is colored by the entire spectrum, while other shorter chains have little or no gradient. If it is not too hard to implement, I'd like to see the default be that each chain is colored from red to blue, regardless of its length or sequence numbering offset. That is what I wish R/C did. ---------------- Actually, there is a bug in R/C such that when less than "all atoms" are selected, the resulting color gradient is generally unsatisfactory. Therefore Protein Explorer suggests a 'select all' prior to applying 'color group' (COLOR "N->C Rainbow" on the QuickViews menu). I hope Jmol will not carry compatibility to the point of emulating this bug :-) /* - - - - - - - - - - - - - - - - - - - - - - - - - - - Protein Explorer - 3D Visualization: http://proteinexplorer.org Workshops: http://www.umass.edu/molvis/workshop Biochem Structure Tutorials http://MolviZ.org World Index of Molecular Visualization Resources: http://molvisindex.org ConSurf - Find Conserved Patches in Proteins: http://consurf.tau.ac.il Atlas of Macromolecules: http://molvis.sdsc.edu/atlas/atlas.htm PDB Lite Macromolecule Finder: http://pdblite.org Molecular Visualization EMail List (molvis-list): http://bioinformatics.org/mailman/listinfo/molvis-list Eric Martz, Professor Emeritus, Dept Microbiology U Mass, Amherst MA US 413-545-2325/FAX 413-545-2532 http://www.umass.edu/molvis/martz - - - - - - - - - - - - - - - - - - - - - - - - - - - */ |
From: Miguel <mi...@jm...> - 2004-11-30 19:15:26
|
Eric wrote: > The main uses of 'color group', in my experience, are > (i) to facilitate tracing the path of a chain with the eye > (ii) to make it easy to see the N and C termini (or 3' and 5' termini).= > > I don't think special treatment for insertion-coded residues would be > useful. For the rare case where it might, the user could write a script= . What I have implemented will handle insertion codes gracefully. > ---------------- > The fact that RasMol/Chime (R/C) use a single gradient, pegged to seque= nce > numbers, for all chains, is unfortunate. Typically, one long chain is > colored by the entire spectrum, while other shorter chains have little = or > no gradient. Yes, that was my observation. When I loaded up 1A00.pdb I was disappointed to see that everything was blue and cyan. I then loaded it up in RasMol 2.6b2 and saw the same thing ... so I guess= 'color group' is compatible. > If it is not too hard to implement, I'd like to see the default be that= > each chain is colored from red to blue, regardless of its length or > sequence numbering offset. That is what I wish R/C did. This can be done. Q: Do you want red->white->blue or blue->white->red? (It seems to me that you want to start with blue and end with red because= you start with a nitrogen and end with an oxygen.) Q: Does it apply to 'chains' (in the PDB sense of the term) or polymers (in the Jmol sense of the term ... groups that are actually connected)? (It seems to me that you only want this to apply to polymers ... any singleton groups/atoms would go to white) Q: What do you want to call it? 'color polymerSequence'? 'color polymer'?= > ---------------- > Actually, there is a bug in R/C such that when less than =22all atoms=22= are > selected, the resulting color gradient is generally unsatisfactory. > Therefore Protein Explorer suggests a 'select all' prior to applying > 'color group' (COLOR =22N->C Rainbow=22 on the QuickViews menu). > I hope Jmol will not > carry compatibility to the point of emulating this bug :-) In what sense is it 'unsatisfactory'? Miguel |
From: Eric M. <em...@mi...> - 2004-11-30 20:42:35
|
At 11/30/04, Miguel wrote: > > If it is not too hard to implement, I'd like to see the default be that > > each chain is colored from red to blue, regardless of its length or > > sequence numbering offset. That is what I wish R/C did. > >This can be done. Great! >Q: Do you want red->white->blue or blue->white->red? > >(It seems to me that you want to start with blue and end with red because >you start with a nitrogen and end with an oxygen.) White in the middle is not very useful in my opinion. I strongly prefer the spectral sequence used in R/C, blue, green, yellow, orange, red. (Magenta is not used in R/C, OK with me -- saves it for something else concurrently.) Consurf uses a 2-hue with white in the middle gradient, and I've never been very happy with the resulting difficulty in distinguishing the middle portions. If 'color group' does the spectral sequence (rainbow), OK. But in that case I'd like 'color group' to act independently on each chain, and on the selected subset, eliminating the bugs in R/C at the cost of compatibility. I don't forsee major compatibility issues. For the most part, people using R/C have had to 'select all' before coloring group in order to get a decent result, so they'll get a better result with existing scripts. In case this hasn't come up, in R/C, one must 'set hetero off' to exclude solvent and ligand from the color assignment before issuing 'color group'. This would not be an issue if each chain gets its own complete blue to red gradient. In R/C, hetero atoms are considered a 'chain', which is not useful in some places (e.g. the chain count is wrong when hetero are present). However, including hetero atoms in color schemes such as 'color chain' and 'color group' makes sense. >Q: Does it apply to 'chains' (in the PDB sense of the term) or polymers >(in the Jmol sense of the term ... groups that are actually connected)? > >(It seems to me that you only want this to apply to polymers ... any >singleton groups/atoms would go to white) Well, solvent molecules and ligands are also given group numbers. This is the main reason for 'group' rather than 'sequence'. I've never found it particularly useful, but I'm not sure it makes sense to exclude them. Especially if we can select polymers (can we?) (else select dna,rna,protein) before applying 'color group', yet get a good result. >Q: What do you want to call it? 'color polymerSequence'? 'color polymer'? if 'color group' is already taken, color sequence? (not strictly applicable to hetero, but OK) > > ---------------- > > Actually, there is a bug in R/C such that when less than "all atoms" are > > selected, the resulting color gradient is generally unsatisfactory. > > Therefore Protein Explorer suggests a 'select all' prior to applying > > 'color group' (COLOR "N->C Rainbow" on the QuickViews menu). > > I hope Jmol will not > > carry compatibility to the point of emulating this bug :-) > >In what sense is it 'unsatisfactory'? Gradient is much steeper -- fewer intermediate shades. Even if you select the single longest chain before 'color group'. |
From: Miguel <mi...@jm...> - 2004-11-30 23:07:07
|
Eric wrote: >> > If it is not too hard to implement, I'd like to see the default be >> that each chain is colored from red to blue, regardless of its length = or >> > sequence numbering offset. =5Bsnip=5D >>Q: Do you want red->white->blue or blue->white->red? > White in the middle is not very useful in my opinion. I strongly prefer= > the spectral sequence used in R/C, blue, green, yellow, orange, red. I misinterpreted your request to go from 'red to blue' ... my mistake. If you want the 'color spectrum' then that is fine. > If 'color group' does the spectral sequence (rainbow), OK. I implemented 'color group' today ... and it is RasMol compatible. > But in that case > I'd like 'color group' to act independently on each chain, and on the > selected subset, eliminating the bugs in R/C at the cost of compatibili= ty. > I don't forsee major compatibility issues. That is exactly where this discussion started ... Do we want to maintain RasMol/Chime compatibility with 'color group' ? > For the most part, people using > R/C have had to 'select all' before coloring group in order to get a > decent result, so they'll get a better result with existing scripts. > > In case this hasn't come up, in R/C, one must 'set hetero off' to exclu= de > solvent and ligand from the color assignment before issuing 'color grou= p'. > This would not be an issue if each chain gets its own complete blue to = red > gradient. In R/C, hetero atoms are considered a 'chain', which is not > useful in some places (e.g. the chain count is wrong when hetero are > present). However, including hetero atoms in color schemes such as 'col= or > chain' and 'color group' makes sense. If we are going to color based upon the selected set then people can pick= the things they want and apply the coloring. >>Q: Does it apply to 'chains' (in the PDB sense of the term) or polymers= >>(in the Jmol sense of the term ... groups that are actually connected)?= > > Well, solvent molecules and ligands are also given group numbers. This = is > the main reason for 'group' rather than 'sequence'. > > I've never found it particularly useful, but I'm not sure it makes sens= e > to > exclude them. Especially if we can select polymers (can we?) Yes ... select polymer > (else select > dna,rna,protein) before applying 'color group', yet get a good result. > > >>Q: What do you want to call it? 'color polymerSequence'? 'color polymer= '? > > if 'color group' is already taken, > color sequence? (not strictly applicable to hetero, but OK) 'color group' is not *taken* It can be changed if there is general agreement that it is the right thin= g to do. I will send out a separate email ... >>In what sense is it 'unsatisfactory'? > > Gradient is much steeper -- fewer intermediate shades. Even if you sele= ct > the single longest chain before 'color group'. That sounds like an artifact of the implementation, probably related (in some obscure way) to the palette limitation of 256 colors. Miguel |
From: Bob H. <ha...@st...> - 2004-11-30 19:25:57
|
Miguel, does "show file" now accept any filename, or is it just the current file? Bob -- Robert M. Hanson, ha...@st..., 507-646-3107 Professor of Chemistry, St. Olaf College 1520 St. Olaf Ave., Northfield, MN 55057 mailto:ha...@st... http://www.stolaf.edu/people/hansonr "Imagination is more important than knowledge." - Albert Einstein |
From: Miguel <mi...@jm...> - 2004-11-30 19:55:42
|
> Miguel, does =22show file=22 now accept any filename, or is it just the= > current file? show file will show the current molecular model show file =22any/path/name=22 will show the file =22../directory/relative/pathname/anyFile.whatever=22= Note that the filename argument must be in double quotation marks. Also note that if you are trying to do this in an applet environment then= you can only access files from the same webserver where the page was loaded from. The pathname should *always* be a directory relative pathname and never start with http:// or / Miguel |