Re: Addition of PDC support to relax 1.3.10 and an inconsistency in the error calculation.Re: Addition of PDC support to relax 1.3.10 and an inconsistency in the error calculation.|
Dear Edward, I'm running a current of output files today and send them. The error of a ratio of numbers y = a/b with given aErr, bErr is yErr = sqrt [ (aErr/a)^2 + (bErr/b)^2], which for a=1 and aErr=0 is yErr = bErr/b Does this correspond to yout notation ? The fit error determination in the PDC depends on the method the user has chosen. Under the tag Fit parameter Error estimation method: the options are - from fit using arbitray y uncertainties - from fit using calculated y uncertainties - from Monte Carlo simulations The first 2 are offered by the our standard Marquardt implementation. The first option assumes the same error for all y values. This then cancales out of the formula, therefore we call it arbitrary. The second option uses the y errors as determined before. As we discussed earlier the base plane RMS is one source for these errors the other is from optional repetition experiments. If repetition experiments are available the systematic error can be calculated as: - worst case per peak scenario - average variance calculation The test output files I have sent so far calculated the fit parameter errors via from fit using arbitray y uncertainties. This option was chosen accidentally. The error there is different from the other 2 methods. Perhaps it is a better idea to generate a new set of output files using the option "from fit using calculated y uncertainties". That usually compares well with MC simulations. In the PDC I do 1000 MC simulations such that a vary the y values based on a gaussian random distribution, the mean is the original y, the width is the error of y. Best regards, Peter On 2/18/2011 4:06 PM, Edward d'Auvergne wrote: Hi Peter, I am now in the process of releasing relax 1.3.10 which will appear on http://nmr-relax.com by next Monday. This version has support for reading the PDC files. If there are additional problems with reading newer versions of the PDC files (>= 1.1.5), then I'll add these files to the test suite, fix the issues, and make a new release. I have also performed some checks to make sure that everything is running ok, and stumbled across an interesting difference between relax and the PDC's error analysis. I have used the peak heights and errors from the R1 data file (testT1.txt version 1.1.5) and re-performed optimisation and Monte Carlo simulations in relax using this. I was trying to make sure that I convert the Tx errors correctly to Rx errors. However there are quite some differences. I have copied and pasted the print out from the script below. As you can see at the end, the PDC/relax ratio for the fitted Rx values are almost the same. However the PDC/relax ratio for the Rx errors vary quite a lot. To compare, I have taken the PDC T1 and T1 errors and used the formula: sigma_R1 = (sigma_T1 / scale) / T1^2. This is then compared to the relax MC error estimates. All of data and scripts are in the directory 'test_suite/shared_data/pdc_files/' in the relax 1.3 main line for reference. Would you know what the issue could be? I think we need to look into this one very carefully. Regards, Edward P. S. Here is the script print out: [edward@localhost pdc_files]$ relax convert_data.py relax repository checkout Molecular dynamics by NMR data analysis Copyright (C) 2001-2006 Edward d'Auvergne Copyright (C) 2006-2011 the relax development team This is free software which you are welcome to modify and redistribute under the conditions of the GNU General Public License (GPL). This program, including all modules, is licensed under the GPL and comes with absolutely no warranty. For details type 'GPL' within the relax prompt. Assistance in using the relax prompt and scripting interface can be accessed by typing 'help' within the prompt. script = 'convert_data.py' ---------------------------------------------------------------------------------------------------- #! /usr/bin/env python # The data - PDC T2, PDC T2 err, PDC scale factor, relax R2, relax R2 err (MC sim). data = [ ['Gln', 2, 0.4560, 0.0055642, 2.22814 , 2.19316068443, 0.0264822775112], ['Ile', 3, 0.4289, 0.0040993, 2.22814 , 2.33150942395, 0.0218282475286], ['Phe', 4, 0.4259, 0.0058593, 2.22814 , 2.34809773828, 0.0393993956573], ['Val', 5, 0.4447, 0.0078459, 2.22814 , 2.24905317514, 0.0190416227806], ['Lys', 6, 0.4388, 0.0068723, 2.22814 , 2.27991249993, 0.0073387122964], ['Thr', 7, 0.4504, 0.0077632, 2.22814 , 2.21988325334, 0.0289980021014], ['Leu', 8, 0.4465, 0.011841 , 2.22814, 2.23992585719, 0.0394637142884], ['Thr', 9, 0.5082, 0.018350 , 2.22814, 1.96827404586, 0.0285585555608], ['Gly', 10, 0.4790, 0.011883 , 2.22814, 2.08825352037, 0.00856203608884], ['Lys', 11, 0.5007, 0.011447 , 2.22814, 1.99766965295, 0.0154821098023], ['Thr', 12, 0.4621, 0.013991 , 2.22814, 2.16418266589, 0.0589894950385], ['Ile', 13, 0.4401, 0.0074423, 2.22814 , 2.27249997894, 0.0323797055027], ['Thr', 14, 0.4548, 0.0067366, 2.22814 , 2.19913359459, 0.0261282530191], ['Leu', 15, 0.4387, 0.0095755, 2.22814 , 2.27947735732, 0.0804008170395], ['Glu', 16, 0.4793, 0.013166 , 2.22814, 2.08663871613, 0.0742976416989], ['Val', 17, 0.4378, 0.0045095, 2.22814 , 2.28454679926, 0.0319478148557], ['Glu', 18, 0.4744, 0.0084630, 2.22814 , 2.10794161516, 0.0491390514111], ['Ser', 20, 0.4459, 0.0078165, 2.22814 , 2.2432376545, 0.0174075720472], ['Asp', 21, 0.4256, 0.0058344, 2.22814 , 2.34905882789, 0.00896916086091], ['Thr', 22, 0.4396, 0.0069851, 2.22814 , 2.27492782819, 0.0283784241729], ['Ile', 23, 0.4054, 0.011480 , 2.22814, 2.46692508258, 0.0512139594604], ['Asn', 25, 0.4175, 0.0053666, 2.22814 , 2.39550908267, 0.028141856637], ['Val', 26, 0.4688, 0.015471 , 2.22814, 2.13297271051, 0.0738836932379], ['Lys', 27, 0.4233, 0.0059289, 2.22814 , 2.36245343286, 0.034473000886], ['Ala', 28, 0.4193, 0.0056136, 2.22814 , 2.38481618154, 0.0369276054641], ['Lys', 29, 0.4298, 0.0061273, 2.22814 , 2.32688199592, 0.0267154381038], ['Thr', 30, 0.4132, 0.010655 , 2.22814, 2.42020223098, 0.0641345037416], ['Gln', 31, 0.4216, 0.0066747, 2.22814 , 2.37198006924, 0.0607157552927], ['Asp', 32, 0.4268, 0.0093311, 2.22814 , 2.343030317, 0.0287089229986], ['Lys', 33, 0.4560, 0.0035919, 2.22814 , 2.19288519999, 0.0322404521468], ['Glu', 34, 0.4468, 0.0039100, 2.22814 , 2.23809730593, 0.016155226898], ['Gly', 35, 0.4452, 0.0042318, 2.22814 , 2.24610605696, 0.0166231784999], ['Ile', 36, 0.5023, 0.0092235, 2.22814 , 1.99097003714, 0.018556395961], ['Asp', 39, 0.4360, 0.0059373, 2.22814 , 2.29368308937, 0.0107946336402], ['Gln', 40, 0.4435, 0.0036092, 2.22814 , 2.25465715695, 0.0219579437623], ['Gln', 41, 0.4496, 0.0036133, 2.22814 , 2.22439489827, 0.0246838738103], ['Arg', 42, 0.4387, 0.016448 , 2.22814, 2.27979894813, 0.14609662846], ['Leu', 43, 0.4603, 0.0091587, 2.22814 , 2.17239195077, 0.0562093749792], ['Ile', 44, 0.4387, 0.0082883, 2.22814 , 2.27944810393, 0.0274015962376], ['Phe', 45, 0.4413, 0.0052304, 2.22814 , 2.26621382042, 0.0325598012054], ['Ala', 46, 0.4604, 0.016900 , 2.22814, 2.17327103743, 0.0366263114163], ['Gly', 47, 0.4564, 0.010048 , 2.22814, 2.19159971705, 0.0200974396044], ['Lys', 48, 0.4543, 0.013460 , 2.22814, 2.20125816954, 0.0769594227925], ['Gln', 49, 0.4731, 0.0092575, 2.22814 , 2.11409184343, 0.0211565921351], ['Leu', 50, 0.4456, 0.0065931, 2.22814 , 2.24441747047, 0.0466054926297], ['Glu', 51, 0.4687, 0.0064085, 2.22814 , 2.13351495913, 0.0369010861993], ['Asp', 52, 0.4888, 0.0082688, 2.22814 , 2.04583349976, 0.0309251585403], ['Arg', 54, 0.4562, 0.0075315, 2.22814 , 2.19228267695, 0.0125176652031], ['Thr', 55, 0.4396, 0.0066617, 2.22814 , 2.27502074947, 0.0319568936121], ['Leu', 56, 0.4221, 0.0050821, 2.22814 , 2.36913113746, 0.0158844775147], ['Ser', 57, 0.4352, 0.0045934, 2.22814 , 2.29814478201, 0.0142640516301], ['Asp', 58, 0.4207, 0.0047524, 2.22814 , 2.37720483191, 0.0272911357828], ['Tyr', 59, 0.4493, 0.0054211, 2.22814 , 2.22553611629, 0.0125933167265], ['Asn', 60, 0.4377, 0.0042069, 2.22814 , 2.28459962433, 0.013370924242], ['Ile', 61, 0.4270, 0.0066972, 2.22814 , 2.34213728573, 0.0264153868176], ['Gln', 62, 0.5067, 0.0066460, 2.22814 , 1.97367886288, 0.0217945020797], ['Lys', 63, 0.4619, 0.0052284, 2.22814 , 2.16500024708, 0.0195312534223], ['Glu', 64, 0.4312, 0.0037115, 2.22814 , 2.31913930394, 0.0223294049604], ['Ser', 65, 0.4568, 0.0047875, 2.22814 , 2.18918681844, 0.00908363177241], ['Thr', 66, 0.4585, 0.0079772, 2.22814 , 2.18167008464, 0.0112902195625], ['Leu', 67, 0.4410, 0.012010 , 2.22814, 2.26772345178, 0.0444564014823], ['His', 68, 0.4452, 0.0033200, 2.22814 , 2.24680733455, 0.00926065649886], ['Leu', 69, 0.4447, 0.0061155, 2.22814 , 2.24851949133, 0.0379125750722], ['Val', 70, 0.4319, 0.0049952, 2.22814 , 2.31543911345, 0.0279226422469], ['Leu', 71, 0.4596, 0.0071340, 2.22814 , 2.1756894166, 0.00987045835498], ['Arg', 72, 0.4622, 0.0058587, 2.22814 , 2.16361975928, 0.0253849160526], ['Leu', 73, 0.5386, 0.013172 , 2.22814, 1.85657597782, 0.0255460422485], ['Arg', 74, 0.6203, 0.026077 , 2.22814, 1.6122677679, 0.0231001217925], ['Gly', 75, 0.8604, 0.062462 , 2.22814, 1.16243924266, 0.023448186193], ['Gly', 76, 1.310, 0.026787 , 2.22814, 0.763064480891, 0.0030557315479]] r1 = [] r1_err = [] print("%-4s %-3s %-15s %-15s %-15s %-15s" % ('Name', 'Num', 'r1', 'err', 'r1_relax_ratio', 'err_relax_ratio')) for i in range(len(data)): r1.append(1.0/data[i][2]) r1_err.append((data[i][3] / data[i][4]) / data[i][2]**2) print("%-4s %-3s %15.10f %15.10f %15.10f %15.10f" % (data[i][0], data[i][1], r1[-1], r1_err[-1], r1[-1]/data[i][5], r1_err[-1]/data[i][6])) print("\nr1 = %s" % r1) print("\nr1_err = %s" % r1_err) ---------------------------------------------------------------------------------------------------- Name Num r1 err r1_relax_ratio err_relax_ratio Gln 2 2.1929824561 0.0120096561 0.9999187345 0.4534978569 Ile 3 2.3315458149 0.0100012696 1.0000156083 0.4581801445 Phe 4 2.3479690068 0.0144973285 0.9999451763 0.3679581436 Val 5 2.2487069935 0.0178059863 0.9998460767 0.9351086556 Lys 6 2.2789425706 0.0160186679 0.9995745761 2.1827627541 Thr 7 2.2202486679 0.0171751894 1.0001646098 0.5922887162 Leu 8 2.2396416573 0.0266564910 0.9998731209 0.6754683762 Thr 9 1.9677292405 0.0318877790 0.9997232065 1.1165753452 Gly 10 2.0876826722 0.0232440925 0.9997266385 2.7147856256 Lys 11 1.9972039145 0.0204924566 0.9997668591 1.3236217091 Thr 12 2.1640337589 0.0294058911 0.9999311948 0.4984936905 Ile 13 2.2722108612 0.0172449508 0.9998727755 0.5325851656 Thr 14 2.1987686895 0.0146169706 0.9998340687 0.5594316083 Leu 15 2.2794620470 0.0223297426 0.9999932834 0.2777302944 Glu 16 2.0863759649 0.0257215134 0.9998740792 0.3461955559 Val 17 2.2841480128 0.0105592823 0.9998254418 0.3305165742 Glu 18 2.1079258010 0.0168768933 0.9999924978 0.3434517522 Ser 20 2.2426553039 0.0176439132 0.9997403973 1.0135769158 Asp 21 2.3496240602 0.0144560781 1.0002406207 1.6117536917 Thr 22 2.2747952684 0.0162223880 0.9999417301 0.5716451318 Ile 23 2.4666995560 0.0313495942 0.9999085799 0.6121298679 Asn 25 2.3952095808 0.0138179556 0.9998749736 0.4910108005 Val 26 2.1331058020 0.0315937156 1.0000623972 0.4276141899 Lys 27 2.3623907394 0.0148502961 0.9999734626 0.4307804867 Ala 28 2.3849272597 0.0143301014 1.0000465772 0.3880593182 Lys 29 2.3266635644 0.0148865427 0.9999061270 0.5572262233 Thr 30 2.4201355276 0.0280085324 0.9999724389 0.4367155083 Gln 31 2.3719165085 0.0168534211 0.9999732035 0.2775790398 Asp 32 2.3430178069 0.0229901363 0.9999946607 0.8008010732 Lys 33 2.1929824561 0.0077526839 1.0000443508 0.2404644904 Glu 34 2.2381378693 0.0087903861 1.0000181240 0.5441202529 Gly 35 2.2461814915 0.0095823570 1.0000335846 0.5764455346 Ile 36 1.9908421262 0.0164069147 0.9999357545 0.8841649395 Asp 39 2.2935779817 0.0140175960 0.9999541751 1.2985707981 Gln 40 2.2547914318 0.0082353341 1.0000595544 0.3750503341 Gln 41 2.2241992883 0.0080224855 0.9999120615 0.3250091764 Arg 42 2.2794620470 0.0383561805 0.9998522233 0.2625398058 Leu 43 2.1724961981 0.0194003430 1.0000479874 0.3451442578 Ile 44 2.2794620470 0.0193280357 1.0000061168 0.7053616691 Phe 45 2.2660321777 0.0120538164 0.9999198475 0.3702054675 Ala 46 2.1720243267 0.0357827405 0.9994263436 0.9769681729 Gly 47 2.1910604733 0.0216494007 0.9997539497 1.0772218339 Lys 48 2.2011886419 0.0292696218 0.9999684146 0.3803253812 Gln 49 2.1137180300 0.0185628797 0.9998231801 0.8774040520 Leu 50 2.2441651706 0.0149024208 0.9998875878 0.3197567495 Glu 51 2.1335609132 0.0130925429 1.0000215391 0.3548010173 Asp 52 2.0458265139 0.0155323661 0.9999965853 0.5022566356 Arg 54 2.1920210434 0.0162415863 0.9998806570 1.2974932631 Thr 55 2.2747952684 0.0154713150 0.9999008884 0.4841307536 Leu 56 2.3691068467 0.0128017702 0.9999897470 0.8059295738 Ser 57 2.2977941176 0.0108846387 0.9998474142 0.7630818348 Asp 58 2.3769907297 0.0120510666 0.9999099353 0.4415743901 Tyr 59 2.2256843980 0.0120523603 1.0000666274 0.9570441654 Asn 60 2.2846698652 0.0098552267 1.0000307454 0.7370639871 Ile 61 2.3419203747 0.0164852312 0.9999073876 0.6240768439 Gln 62 1.9735543714 0.0116175902 0.9999369242 0.5330514163 Lys 63 2.1649707729 0.0109984227 0.9999863861 0.5631191428 Glu 64 2.3191094620 0.0089587927 0.9999871323 0.4012105415 Ser 65 2.1891418564 0.0102970808 0.9999794618 1.1335863312 Thr 66 2.1810250818 0.0170305756 0.9997043536 1.5084361698 Leu 67 2.2675736961 0.0277155405 0.9999339621 0.6234319383 His 68 2.2461814915 0.0075177053 0.9997214523 0.8117896720 Leu 69 2.2487069935 0.0138789061 1.0000833892 0.3660765864 Val 70 2.3153507756 0.0120183265 0.9999618484 0.4304150871 Leu 71 2.1758050479 0.0151576052 1.0000531470 1.5356536249 Arg 72 2.1635655560 0.0123083235 0.9999749479 0.4848676073 Leu 73 1.8566654289 0.0203787034 1.0000481807 0.7977244855 Arg 74 1.6121231662 0.0304166646 0.9999103116 1.3167317865 Gly 75 1.1622501162 0.0378680208 0.9998373021 1.6149658876 Gly 76 0.7633587786 0.0070054980 1.0003856787 2.2925763776 . --
| |||||||||