### Table 6 Measured and Theoretical Delay Structure Functions

"... In PAGE 28: ...tions, based upon the model of qleuhaft and Ijanyi (1987), and the results are presented in Table6 . The parameters in the model include elevation angle (chosen to be the same as in the individual VLBI scans), vector wind veloc- ity, tropospheric slab height (chosen to be 2 km, the mean wet tropospheric scale height), and the structure corwtant Cx (7 x 10-s m-1/3), Vector wind velocities as a function of height were measured by optical tracking of our ra- diosondes at DSS 13, with launches every 6 hours, The vector wind velocities used in the structure function calculations were linear interpolations (in time) of the values measured by radiosondc tracking.... In PAGE 28: ... The parameters in the model include elevation angle (chosen to be the same as in the individual VLBI scans), vector wind veloc- ity, tropospheric slab height (chosen to be 2 km, the mean wet tropospheric scale height), and the structure corwtant Cx (7 x 10-s m-1/3), Vector wind velocities as a function of height were measured by optical tracking of our ra- diosondes at DSS 13, with launches every 6 hours, The vector wind velocities used in the structure function calculations were linear interpolations (in time) of the values measured by radiosondc tracking. For each scan, the square root of the ratio of measured to theoretical value of D~ is given in column 4 of Table6 . The value of At used to determine this ratio was 50s for 700-800s scans and 16 s for 10 amp;150s scans, as indicated in column 3.... In PAGE 30: ...A comparison of VLBI-measured and the model-predicted interscan residuals is given in Table 7. For each sessioll, the predicted delay rrns due to the troposphel e, based on the VLB1 intnmcan structure functions (ccdumn 6 of Table6 ) is shown in column 2. The Ineasured VLB1 rms, with a solution for a mean troposphere, but without WVR calibration (column 2 of Table 4) is given in column 3.... ..."

### TABLE 2. Categorized items from the first part of the model perception questionnaire.

### Table 1. Definitions of the function types, their effects and predicted risks of SNPs

2006

"... In PAGE 2: ... A high risk rank implies a high-risk level. Table1 gives the definitions of the function types, effects and their predicted risk ranking. For a coding SNP, if it is non-synonymous and alters an amino acid in a protein resulting in a different protein structure (mis-sense, non-conservative change), or a non-sense change that results in a premature termination of the amino acid sequence (non-sense), then it will be assigned a high-risk or very high-risk ranking, because most known disease-causing SNPs are in these classes.... ..."

### Table 3: Predicted function/contacts of conserved residues. Homologous residues making similar contacts in different structures are in the same rows. Interacting functional groups are indicated in parentheses ( quot;b quot; denotes backbone). Contacts absent from the lowest- energy solution, but present in other quot;top 20 quot; solutions are shown in italics.

"... In PAGE 5: ... Despite similar conformations, the protein-ligand interac- tions in the CbiF, SET and SPOUT superfamilies are differ- ent. The key interactions between the three SPOUT MTases and AdoMet, present in the lowest-energy docking solution and in the majority of sub-optimal solutions, are listed in Table3 a. It is striking that despite the structural variability and extensive sequence divergence, many pro- tein-ligand interactions were found to be conserved in all three cases.... ..."

### Table 1: Multiple regression analyses that predict LSA match scores

"... In PAGE 5: ...67). Table1 presents the results of the multiple regression analysis that predicted LSA match scores as a function of the three structural proximity scores, length, and noun overlap. The five predictor variables accounted for a significant 55% of the variance in LSA match scores, , F(5, 530) = 128.... ..."

### Table 5. Correlation Coefficients for Aqueous Solubility Blind Test Regressionsa

1998

"... In PAGE 9: ... Second, the calculated coefficients of the descrip- tors for the two group regressions are all within the error estimate of the coefficients for the regression made with all structures, suggesting that the coefficient values are reliable. Finally, when considering the correlation coefficients of the predictions for the blind tests in Table5 , it is apparent that the ability of the regressions made using two-thirds of the structures to predict the aqueous solubility for the excluded third is essentially equal. The average correlation coefficient for the blind cases (AB f C, AC f B, BC f A) was equal to the correlation coefficient (R2 ) 0.... ..."

Cited by 2

### (Table 2). The fraction of correct assignments is similar for all four experiments, indicating that our estimates have low variance and high confidence. There is no obvious increase in the fractions of correct assignments for the three consecutive LiveBench experiments, indicating that increasing quality in structure prediction may not neces- sarily lead to improvements in structure-based annotation transfer. Overall, we found that approximately one-third

2001

"... In PAGE 7: ... The primary differ- ence between the two types of experiments is that Live- Bench uses proteins with newly deposited structures in the Protein Data Bank (PDB) [42] as targets, while PDB- CAFASP collects pre-released sequences (usually weeks before the experimental structures are released) in the PDB as targets. Functional assignments based on predicted structural similarity To investigate the correlation between successful fold rec- ognition and correct functional assignment, we analyzed hard prediction targets collected from the LiveBench 7, Page 7 of 10 (page number not for citation purposes) fies enzyme function by computing the theoretical micro- scopic titration curve for each residue in a protein LiveBench 8 and LiveBench 9 and PDB-CAFASP 1 experi- ments (Table2... ..."

Cited by 1

### Table 5: Predictive Validity Structural

"... In PAGE 28: ... Predictive Validity We assess predictive validity by examining whether the distinction between the five configurations is useful in predicting differences along other quot;dependent quot; variables -- reflecting the performance of the relations. Table5... ..."

### Table 5. Learning techniques

2003

"... In PAGE 22: ...nderlying technologies, e.g., support vector machines [Cristianini and Shawe-Taylor, 2000]; improved links between reinforcement learning and stochastic control theory [Bertsekas and Tsitsiklis, 1996]; advances in planning and learning methods for stochastic environments [Littamn, 1996; Parr, 1998]; and improved theoretical models of simple genetic algorithms [Vose, 1999]. Major types of learning techniques are summarized in Table5 [Zimmerman and Kambhampati, 2001; Nordlander, 2001]. ... ..."

Cited by 2

### Table 1: Comparison between the Inverse-SF scaling exponents syn(p) measured in the synthetic signal and the inversion of the theoretical multifractal prediction (3.6), th(p). The synthetic signal has been de ned such has the D(h) function leads to the same set of experimental (p) exponents for the direct structure functions.

"... In PAGE 13: ...s in agreement with the prediction (3.6). The same agreement also holds for higher moments. In Table1 , we compare the best t to the p( v) measured on the synthetic eld with the inversion formula (3.6).... In PAGE 13: ...6) is h = 0:5. Therefore, the theoretical prediction, th(q), in Table1 has been obtained by imposing hmax = 0:5. Let us now go back to the most interesting question about the statistical properties of the IDR.... ..."