1. How do I create peak lists for AutoAssign?
2. How do I create an AutoAssign control (.aat) file?
3. How should I order the dimensions in a peak (.pks) file?
4. How do I indicate that the sample is partially or fully deuterated?
5. What does a "-Xmx128m: illegal argument" error mean?
6. What does an "Average value observed for ..." error mean?
7. What does a "While initializing data - no reference peaks defined (for spectrum with 'ROOT' as reference.)" error mean?
8. What is the first step I should do once I have my peak lists in AutoAssign format?
9. Why am I getting an "unable to connect to server" error after several seconds of running "Refined Execution"?
10. How do tolerances affect the performance of AutoAssign?
11. How does processed digital resolution relate to peak picking and and peak list alignment?
12. How do I use the calculate_registration program?
13. How do I increase the resolution of the GIF file created by the CMap Editor?
14. The AutoAssign windows are not working properly under KDE window manager. What do I do?
15. Why are some of my cysteines not being assigned?
1. How do I create peak lists for AutoAssign?Use the extract_columns.pl and create_peak_list.pl AutoPeak perl scripts. Every spectral processing and peak picking program will have a peak format that has a peak per line with the different dimensions and intensity of the peak in designated columns.An example would be a raw peak file (HNCACB) in the following format:
2 126.257 46.356 7.965 2 U 3.396e+04 4.86e+02 a 0 30103 30105 30104 0The following series of commands will create a peak list in the AutoAssign format:
4 124.691 61.261 9.144 2 U 4.550e+02 3.54e+02 a 0 30203 30205 30204 0
5 124.050 54.715 7.640 2 U 6.764e+03 5.84e+02 a 0 30303 30305 30304 0
6 123.694 53.532 8.140 2 U 2.642e+04 4.75e+02 a 0 30403 30405 30404 0
7 123.694 59.841 7.878 2 U 7.815e+03 5.44e+02 a 0 30503 30505 30504 0
...% extract_columns.pl 4 2 3 7 < raw_peak_list.txt > extracted_peak_list.txtThe resulting HNCACB peak file will look like this:
% create_peak_list.pl extracted_peak_list.txt HNCACB hncacb.pks#Index 1Dim 2Dim 3Dim Intensity WorkbookSolution Author: Hunter Moseley
1 7.965 126.257 46.356 3.396e+04 HNCACB
2 9.144 124.691 61.261 4.550e+02 HNCACB
3 7.640 124.050 54.715 6.764e+03 HNCACB
4 8.140 123.694 53.532 2.642e+04 HNCACB
5 7.878 123.694 59.841 7.815e+03 HNCACB
...
2. How do I create an AutoAssign control (.aat) file?
The best suggestion is to copy a table file from one of the example datasets located in the ExampleData subdirectory of the AutoAssign distribution. Then modify this table file to suit your needs.Solution Author: Hunter Moseley
3. How should I order the dimensions in a peak (.pks) file?
They should be in the same order as indicated in the table file. With this said, it is preferable that you use an order of "HN N15 other" in your table file and peak files. This ordering is easier to work with when using many of the AutoPeak scripts.Solution Author: Hunter Moseley
4. How do I indicate that the sample is partially or fully deuterated?
Use the keyword "deuterated" in the Properties section of the table file to specify a fully deuterated protein. Use the keyword "deuterated_%%" to specify a partially deuterated protein of %% percent deuteration.
Solution Author: Hunter Moseley
5. What does a "-Xmx128m: illegal argument" error mean?
This error occurs when the AutoAssign startup scripts point to a pre Java 2.0 (1.2) virtual machine. You should first find out where a Java 2.0 (1.2.?) or Java 3.0 (1.3.?) virtual machine is installed and then reinstall AutoAssign giving it the correct location for Java 2.0 or greater. You may have to install Java 2.0 or greater on the machine first. This is quite common for SGI computers running IRIX. It is common on the Linux machines for Kaffe to be the default Java virtual machine which is not Java 2.0 compliant.
Solution Author: Hunter Moseley
6. What does an "Average value observed for ..." error mean?
This error normally indicates that the order of dimension in your peak list does not match the order specified in the table file. Go back and check the order of the peak dimensions in both files.
Solution Author: Hunter Moseley
7. What does a "While initializing data - no reference peaks defined (for spectrum with 'ROOT' as reference.)" error mean?
This error is occurs when AutoAssign cannot find even a few isolated (at least one) isolated peak in the amide plane across all the spectra to reference by. This usually means that one of the peak lists files is of poor quality or shifted with respect to the other peak list files. Use the calculate_shifting.pl perl script or calculate_registration program to evaluate the shifting between the HSQC peak list and all the other peak lists in the dataset.
Solution Author: Hunter Moseley
8. What is the first step I should do once I have my peak lists in AutoAssign format?
You should use the calculate_registration program to check how well aligned the peak lists are to each other. A typical scheme is to compare the H and N dimensions of each peak list to the HSQC peak list. Then compare all three dimensions of the HNCA, HNcoCACB, and HNcoCA to the HNCACB. Compare all three dimensions of the HNHA to the HNcoHA. Compare all three dimensions of the HNcaCO to the HNCO. Correct any systematic shifts with the apply_shifting.pl perl script. The better aligned the peak lists, the better your results will be from AutoAssign.
Solution Author: Hunter Moseley
9. Why am I getting an "unable to connect to server" error after several seconds of running "Refined Execution"?
This often occurs because no HA tolerance is listed in the table file. Even though you do not have neither an HNHA nor HNcoHA type peak list, "Refined Execution" still calls Incremental Match on HA's which then causes the AutoServer to crash. Simply put a nonzero HA tolerance in the table file and the error should go away.
Solution Author: Hunter Moseley
10. How do tolerances affect the performance of AutoAssign
HN and N15 tolerances affect AutoAssign performance differently than CO, CA, CB, and HA tolerances. HN and N15 tolerances affect the creation of generic spin systems (GS's) and which GS's are considered overlapped. If the HN and N15 tolerances are too tight, the GS's start missing shifts. If the tolerances are too loose, more GS's are considered overlapped. Both situations degrade the performance of AutoAssign. If the CO, CA, CB, and/or HA tolerances are too tight, then AutoAssign does not see all the sequential connectivities (links) between GS's. If these tolerances are too loose, AutoAssign sees too many possible (but improbable) links and cannot decide which ones are real. Either of these situations also degrade performance. In other words, you want the tolerances that match the quality of your peak lists. The better the resolution and alignment of your peak lists, the better the performance from AutoAssign.
Solution Author: Hunter Moseley
11. How does processed digital resolution relate to peak picking and and peak list alignment?
The processed digital resolution of any given dimension of a spectrum is the final digital resolution of that dimension after linear prediction and zero filling. Most automatic peak pickers do not interpolate the peak center. They pick the point with the largest intensity value. Thus for most automatic peak pickers, you are limited to the processed digital resolution of the spectrum. When aligning peak lists from spectra with different processed digital resolutions in comparable dimensions, you will have noncorrectable discrepancies arising from the differences in the processed digital resolutions. It is best that the processed digital resolutions between comparable dimensions match. In a sense, the linear prediction and zero filling turns into the peak center interpolation for the peak picker. Keeping similar digital resolutions for comparable dimensions during acquisition will further improve alignment between peak lists. Ultimately the quality of alignment determines how tight the tolerances have to be.
Solution Author: Hunter Moseley and G.V.T. Swapna
12. How do I use the calculate_registration program?This program compares to peak lists and returns the standard deviations for matching in the dimensions compared. If the robustness score is low, you may have problems with one of the peak lists. You should have significant number of comparable peaks from both peak lists. Multiply the amide dimension standard deviations by 3 to 4 for ballpark tolerances to use inside AutoAssign. Multiply the CA/CB/HA dimension standard deviations by 2 to 3 for ballpark tolerances to use in AutoAssign. Use the worst standard deviations seen in all the registrations for a given dimension.
This program is a replacement for the calculate_shifting.pl perl script and has an improved algorithm for alignment. Look in the AutoPeak Documentation under "Examples of Use" for an example of using this program.
Solution Author: Hunter Moseley and David Snyder
13. How do I increase the resolution of the GIF file created by the CMap Editor ?
Use the "Relative Image Scale" in the "Image Modification Options" to increase the size of the image by several hunded percent. The GIF that is created will have better resolution. The GIF creator is limited to the resolution of the screen which is 72 pixels per inch on most computers.
Solution Author: Hunter Moseley
14. The AutoAssign windows are not working properly under KDE window manager. What do I do?
You have two options: change window managers (Gnome works well) or upgrade the version of Java. Java 1.4 appears to fix the problems with the KDE window manager.
Solution Author: Brian F. Volkman
15. Why are some of my cysteines not being assigned?
You may have nondisulfide-bonded cysteines. They expected values for a "C" amino acid in the sequence in the AutoAssign table file are for a disulfide-bonded cysteine. Use a "B" amino acid to indicate a nondisulfide-bonded cysteine. Use a "U" to indicate a cysteine of unknown oxidation state.
Solution Author: Hunter Moseley