1Hvq Lichtarge Lab 2006

Pages 1–8 1hvq Evolutionary trace report by report maker May 22, 2010

4.3.1 Alistat 7 4.3.2 CE 7 4.3.3 DSSP 7 4.3.4 HSSP 7 4.3.5 LaTex 7 4.3.6 Muscle 7 4.3.7 Pymol 7 4.4 Note about ET Viewer 7 4.5 Citing this work 7 4.6 About report maker 7 4.7 Attachments 7

1 INTRODUCTION From the original Protein Data Bank entry (PDB id 1hvq): Title: Crystal structures of hevamine, a plant defence protein with chitinase and lysozyme activity, and its complex with an inhibitor Compound: Mol id: 1; molecule: hevamine a; chain: a; ec: 3.2.1.14, 3.2.1.17; other details: plant endochitinase/lysozyme Organism, scientiﬁc name: Hevea Brasiliensis; 1hvq contains a single unique chain 1hvqA (273 residues long).

CONTENTS 2 CHAIN 1HVQA 2.1 P23472 overview 1 Introduction 1 From SwissProt, id P23472, 95% identical to 1hvqA: 2 Chain 1hvqA 1 Description: Hevamine A precursor [Includes: Chitinase (EC 2.1 P23472 overview 1 3.2.1.14); Lysozyme (EC 3.2.1.17)]. 2.2 Multiple sequence alignment for 1hvqA 1 Organism, scientiﬁc name: Hevea brasiliensis (Para rubber tree). 2.3 Residue ranking in 1hvqA 1 Taxonomy: Eukaryota; Viridiplantae; Streptophyta; Embryophyta; 2.4 Top ranking residues in 1hvqA and their position on Tracheophyta; Spermatophyta; Magnoliophyta; eudicotyledons; core the structure 2 eudicotyledons; rosids; eurosids I; Malpighiales; Euphorbiaceae; 2.4.1 Clustering of residues at 25% coverage. 2 Crotonoideae; Micrandreae; Hevea. 2.4.2 Overlap with known functional surfaces at Function: Bifunctional enzyme with lysozyme / chitinase activity. 25% coverage. 2 May have a role in plugging the latex vessel and cessation of latex 2.4.3 Possible novel functional surfaces at 25% ﬂow. coverage. 4 Catalytic activity: Random hydrolysis of N-acetyl-beta-D- glucosa- minide 1,4-beta-linkages in chitin and chitodextrins. 3 Notes on using trace results 6 Catalytic activity: Hydrolysis of 1,4-beta-linkages between N- ace- 3.1 Coverage 6 tylmuramic acid and N-acetyl-D-glucosamine residues in a peptido- 3.2 Known substitutions 6 glycan and between N-acetyl-D-glucosamine residues in chitodex- 3.3 Surface 6 trins. 3.4 Number of contacts 6 Subcellular location: In the lutoids (vacuoles) from rubber latex. 3.5 Annotation 6 Miscellaneous: Two components of hevamine have been isolated: 3.6 Mutation suggestions 6 hevamine A (shown here), the most abundant, and hevamine B. Similarity: Belongs to the glycosyl hydrolase 18 family. Chitinase 4 Appendix 6 class II subfamily. 4.1 File formats 6 About: This Swiss-Prot entry is copyright. It is produced through a 4.2 Color schemes used 6 collaboration between the Swiss Institute of Bioinformatics and the 4.3 Credits 7 EMBL outstation - the European Bioinformatics Institute. There are

1 Lichtarge lab 2006 in 1hvqA can be found in the ﬁle called 1hvqA.ranks sorted in the attachment. 2.4 Top ranking residues in 1hvqA and their position on the structure In the following we consider residues ranking among top 25% of residues in the protein . Figure 3 shows residues in 1hvqA colored by their importance: bright red and yellow indicate more conser- Fig. 1. Residues 1-136 in 1hvqA colored by their relative importance. (See ved/important residues (see Appendix for the coloring scheme). A Appendix, Fig.9, for the coloring scheme.) Pymol script for producing this ﬁgure can be found in the attachment.

Fig. 2. Residues 137-273 in 1hvqA colored by their relative importance. (See Appendix, Fig.9, for the coloring scheme.) no restrictions on its use as long as its content is in no way modiﬁed and this statement is not removed.

2.2 Multiple sequence alignment for 1hvqA For the chain 1hvqA, the alignment 1hvqA.msf (attached) with 217 sequences was used. The alignment was downloaded from the HSSP database, and fragments shorter than 75% of the query as well as duplicate sequences were removed. It can be found in the attachment to this report, under the name of 1hvqA.msf. Its statistics, from the alistat program are the following: Fig. 3. Residues in 1hvqA, colored by their relative importance. Clockwise: Format: MSF front, back, top and bottom views. Number of sequences: 217 Total number of residues: 56932 Smallest: 214 2.4.1 Clustering of residues at 25% coverage. Fig. 4 shows the Largest: 273 top 25% of all residues, this time colored according to clusters they Average length: 262.4 belong to. The clusters in Fig.4 are composed of the residues listed Alignment length: 273 in Table 1. Average identity: 39% Most related pair: 99% Table 1. Most unrelated pair: 15% cluster size member Most distant seq: 36% color residues red 64 6,7,8,16,32,36,45,57,63,64 67,68,71,72,74,76,77,78,79 Furthermore, <1% of residues show as conserved in this ali- 80,93,97,101,104,112,113,114 gnment. 115,117,119,120,121,123,124 The alignment consists of 47% eukaryotic ( 13% fungi, 34% 125,126,127,154,155,156,157 plantae), and <1% prokaryotic sequences. (Descriptions of some 158,159,162,163,169,170,175 sequences were not readily available.) The ﬁle containing the 176,178,179,180,181,182,183 sequence descriptions can be found in the attachment, under the name 184,185,217,219,220,228,229 1hvqA.descr. 253,255 continued in next column 2.3 Residue ranking in 1hvqA The 1hvqA sequence is shown in Figs. 1–2, with each residue colored according to its estimated importance. The full listing of residues

2 Table 2. continued res type subst’s cvg noc/ dist (%) bb (A˚ ) R(14)H A(2) .(4)LTG EMFKVN

Table 2. The top 25% of residues in 1hvqA at the interface with NAG.(Field names: res: residue number in the PDB entry; type: amino acid type; substs: substitutions seen in the alignment; with the percentage of each type in the bracket; noc/bb: number of contacts with the ligand, with the number of contacts realized through backbone atoms given in the bracket; dist: distance of closest apporach to the ligand. )

Table 3. res type disruptive mutations 80 G (FW)(HR)(Y)(KE) 255 W (E)(K)(D)(T) 32 F (K)(E)(Q)(D) Fig. 4. Residues in 1hvqA, colored according to the cluster they belong to: 9 Q (Y)(T)(H)(FW) red, followed by blue and yellow are the largest clusters (see Appendix for the coloring scheme). Clockwise: front, back, top and bottom views. The Table 3. List of disruptive mutations for the top 25% of residues in corresponding Pymol script is attached. 1hvqA, that are at the interface with NAG.

Table 1. continued cluster size member color residues blue 3 249,250,251

Table 1. Clusters of top ranking residues in 1hvqA.

2.4.2 Overlap with known functional surfaces at 25% coverage. The name of the ligand is composed of the source PDB identiﬁer and the heteroatom name used in that ﬁle. NAG binding site. Table 2 lists the top 25% of residues at the interface with 1hvqNAG3 (nag). The following table (Table 3) suggests possible disruptive replacements for these residues (see Section 3.6). Table 2. res type subst’s cvg noc/ dist (%) bb (A˚ ) 80 G G(99)KE 0.01 3/3 3.94 255 W W(94) 0.05 24/0 3.23 Y(2)RLF V. 32 F F(93) 0.12 5/0 3.99 .(1) Fig. 5. Residues in 1hvqA, at the interface with NAG, colored by their relative importance. The ligand (NAG) is colored green. Atoms further than 30A˚ away Y(1)A from the geometric center of the ligand, as well as on the line of sight to the V(1)TGL ligand were removed. (See Appendix for the coloring scheme for the protein Q chain 1hvqA.) 9 Q Q(72)P 0.25 16/1 3.41 continued in next column Figure 5 shows residues in 1hvqA colored by their importance, at the interface with 1hvqNAG3.

3 NAG binding site. Table 4 lists the top 25% of residues at the interface with 1hvqNAG1 (nag). The following table (Table 5) suggests possible disruptive replacements for these residues (see Section 3.6). Table 4. res type subst’s cvg noc/ dist (%) bb (A˚ ) 45 N N(66) 0.22 5/0 3.53 D(22) .(2) S(2) A(1) G(2)VTE 9 Q Q(72)P 0.25 1/1 4.83 R(14)H A(2) .(4)LTG EMFKVN

Table 4. The top 25% of residues in 1hvqA at the interface with NAG.(Field names: res: residue number in the PDB entry; type: amino acid type; substs: substitutions seen in the alignment; with the percentage of each Fig. 6. Residues in 1hvqA, at the interface with NAG, colored by their relative type in the bracket; noc/bb: number of contacts with the ligand, with the num- importance. The ligand (NAG) is colored green. Atoms further than 30A˚ away ber of contacts realized through backbone atoms given in the bracket; dist: from the geometric center of the ligand, as well as on the line of sight to the distance of closest apporach to the ligand. ) ligand were removed. (See Appendix for the coloring scheme for the protein chain 1hvqA.)

Table 5. res type disruptive Table 6. continued mutations res type subst’s cvg noc/ dist 45 N (Y)(H)(FW)(R) (%) bb (A˚ ) 9 Q (Y)(T)(H)(FW) EMFKVN

Table 5. List of disruptive mutations for the top 25% of residues in Table 6. The top 25% of residues in 1hvqA at the interface with 1hvqA, that are at the interface with NAG. NAG.(Field names: res: residue number in the PDB entry; type: amino acid type; substs: substitutions seen in the alignment; with the percentage of each Figure 6 shows residues in 1hvqA colored by their importance, at the type in the bracket; noc/bb: number of contacts with the ligand, with the number of contacts realized through backbone atoms given in the bracket; dist: interface with 1hvqNAG1. distance of closest apporach to the ligand. ) NAG binding site. Table 6 lists the top 25% of residues at the interface with 1hvqNAG2 (nag). The following table (Table 7) suggests possible disruptive replacements for these residues (see Section Table 7. 3.6). res type disruptive Table 6. mutations res type subst’s cvg noc/ dist 45 N (Y)(H)(FW)(R) (%) bb (A˚ ) 9 Q (Y)(T)(H)(FW) 45 N N(66) 0.22 23/0 3.25 D(22) Table 7. List of disruptive mutations for the top 25% of residues in .(2) 1hvqA, that are at the interface with NAG. S(2) A(1) Figure 7 shows residues in 1hvqA colored by their importance, at the G(2)VTE interface with 1hvqNAG2. 9 Q Q(72)P 0.25 29/8 2.83 R(14)H 2.4.3 Possible novel functional surfaces at 25% coverage. One A(2) group of residues is conserved on the 1hvqA surface, away from (or .(4)LTG susbtantially larger than) other functional sites and interfaces reco- continued in next column gnizable in PDB entry 1hvq. It is shown in Fig. 8. The right panel shows (in blue) the rest of the larger cluster this surface belongs to. The residues belonging to this surface ”patch” are listed in Table

4 Table 8. continued res type substitutions(%) cvg 255 W W(94)Y(2)RLFV. 0.05 127 E E(82)D(12)Q(1)K 0.06 GA.LV 120 D D(92)N(5)LEV. 0.07 184 N N(83)D(5)G(7)SY 0.07 EAK 68 Q Q(94).R(1)K(2) 0.09 H(1)L 162 P P(91)VQNEGD(1)M 0.10 LFTSK 125 D N(5)D(75)F(12) 0.11 Y(2)WEV(1).HI 32 F F(93).(1)Y(1)A 0.12 V(1)TGLQ 185 N N(76)D(12)G(1) 0.13 S(2)T(3)AKE(1)R 7 W W(89)V.(5)Y(3)G 0.14 R 71 G G(84)DN(11)YARH 0.15 I Fig. 7. Residues in 1hvqA, at the interface with NAG, colored by their relative 112 R R(90)G(5)WKAPHS 0.15 importance. The ligand (NAG) is colored green. Atoms further than 30A˚ away . from the geometric center of the ligand, as well as on the line of sight to the 115 G G(88)TD(7)R(1)A 0.16 ligand were removed. (See Appendix for the coloring scheme for the protein EY. chain 1hvqA.) 228 G G(84)D(5)A(4) 0.16 Q(2)HN.VK 101 W W(86)Y(6)F(3) 0.17 .(1)VLH 16 L L(88)S(2)A.(3) 0.18 N(1)VGTI 114 L L(41)F(55)ISGAM 0.18 . 63 G Q(11)G(4)D(65) 0.19 E(10).A(4)NS(2) 72 I I(35)K(33)V(27) 0.19 T.NRQ 119 L L(52)V(35)I(7) 0.21 Fig. 8. A possible active surface on the chain 1hvqA. The larger cluster it F(1)QM(1). belongs to is shown in blue. 45 N N(66)D(22).(2) 0.22 S(2)A(1)G(2)VTE 104 F F(71)V(3)Y(20) 0.22 8, while Table 9 suggests possible disruptive replacements for these .(2)H(1)LM residues (see Section 3.6). 117 A A(80)D(1)V(9) 0.23 Y(1)T(1)EN(1)LP Table 8. I.S res type substitutions(%) cvg 229 Y Y(62)M(2)F(24) 0.23 163 D D(92)Q(1)E(3)S 0.03 W(4)SLRHN.VTA P(1)N 9 Q Q(72)PR(14)H 0.25 183 Y Y(92)F(6)DR 0.03 A(2).(4)LTGEMFK 158 Q Q(80)R(13)E(1)G 0.04 VN LKSA. 181 Q Q(82)R(13)GK(1) 0.04 Table 8. Residues forming surface ”patch” in 1hvqA. END 113 P P(96)IVEF.L 0.05 continued in next column

5 Table 9. one may try replacing, R with K, or (perhaps more surprisingly), with res type disruptive V. The percentage of times the substitution appears in the alignment mutations is given in the immediately following bracket. No percentage is given 163 D (R)(H)(FW)(Y) in the cases when it is smaller than 1%. This is meant to be a rough 183 Y (K)(QM)(EVA)(NLPIR) guide - due to rounding errors these percentages often do not add up 158 Q (Y)(FWH)(T)(CG) to 100%. 181 Q (Y)(FW)(H)(T) 113 P (R)(Y)(TH)(K) 255 W (E)(K)(D)(T) 3.3 Surface 127 E (H)(FW)(Y)(R) To detect candidates for novel functional interfaces, first we look for 120 D (R)(H)(FW)(Y) residues that are solvent accessible (according to DSSP program) by 2 184 N (Y)(FWH)(R)(T) at least 10A˚ , which is roughly the area needed for one water mole- 68 Q (Y)(T)(FW)(S) cule to come in the contact with the residue. Furthermore, we require 162 P (Y)(R)(H)(T) that these residues form a “cluster” of residues which have neighbor 125 D (R)(K)(H)(CG) within 5A˚ from any of their heavy atoms. 32 F (K)(E)(Q)(D) Note, however, that, if our picture of protein evolution is correct, 185 N (Y)(FW)(H)(R) the neighboring residues which are not surface accessible might be 7 W (E)(K)(D)(Q) equally important in maintaining the interaction specificity - they 71 G (E)(KR)(QMD)(FWH) should not be automatically dropped from consideration when choo- 112 R (D)(T)(E)(Y) sing the set for mutagenesis. (Especially if they form a cluster with 115 G (KR)(E)(FWH)(QM) the surface residues.) 228 G (E)(R)(K)(FWH) 101 W (K)(E)(Q)(D) 16 L (R)(Y)(H)(K) 3.4 Number of contacts 114 L (R)(Y)(H)(TK) Another column worth noting is denoted “noc/bb”; it tells the num- 63 G (R)(H)(FKW)(E) ber of contacts heavy atoms of the residue in question make across 72 I (Y)(HR)(T)(E) the interface, as well as how many of them are realized through the 119 L (Y)(R)(T)(H) backbone atoms (if all or most contacts are through the backbone, 45 N (Y)(H)(FW)(R) mutation presumably won’t have strong impact). Two heavy atoms 104 F (K)(E)(T)(Q) are considered to be “in contact” if their centers are closer than 5A˚ . 117 A (R)(K)(Y)(H) 229 Y (K)(Q)(E)(M) 9 Q (Y)(T)(H)(FW) 3.5 Annotation If the residue annotation is available (either from the pdb file or Table 9. Disruptive mutations for the surface patch in 1hvqA. from other sources), another column, with the header “annotation” appears. Annotations carried over from PDB are the following: site (indicating existence of related site record in PDB ), S-S (disulfide bond forming residue), hb (hydrogen bond forming residue, jb (james 3 NOTES ON USING TRACE RESULTS bond forming residue), and sb (for salt bridge forming residue). 3.1 Coverage Trace results are commonly expressed in terms of coverage: the residue is important if its “coverage” is small - that is if it belongs to 3.6 Mutation suggestions some small top percentage of residues [100% is all of the residues Mutation suggestions are completely heuristic and based on comple- in a chain], according to trace. The ET results are presented in the mentarity with the substitutions found in the alignment. Note that form of a table, usually limited to top 25% percent of residues (or they are meant to be disruptive to the interaction of the protein to some nearby percentage), sorted by the strength of the presumed with its ligand. The attempt is made to complement the following evolutionary pressure. (I.e., the smaller the coverage, the stronger the properties: small [AV GSTC], medium [LPNQDEMIK], large pressure on the residue.) Starting from the top of that list, mutating a [WFYHR], hydrophobic [LPVAMWFI], polar [GTCY ]; posi- couple of residues should affect the protein somehow, with the exact tively [KHR], or negatively [DE] charged, aromatic [WFYH], effects to be determined experimentally. long aliphatic chain [EKRQM], OH-group possession [SDETY ], and NH2 group possession [NQRK]. The suggestions are listed 3.2 Known substitutions according to how different they appear to be from the original amino One of the table columns is “substitutions” - other amino acid types acid, and they are grouped in round brackets if they appear equally seen at the same position in the alignment. These amino acid types disruptive. From left to right, each bracketed group of amino acid may be interchangeable at that position in the protein, so if one wants types resembles more strongly the original (i.e. is, presumably, less to affect the protein by a point mutation, they should be avoided. For disruptive) These suggestions are tentative - they might prove disrup- example if the substitutions are “RVK” and the original protein has tive to the fold rather than to the interaction. Many researcher will an R at that position, it is advisable to try anything, but RVK. Conver- choose, however, the straightforward alanine mutations, especially in sely, when looking for substitutions which will not affect the protein, the beginning stages of their investigation.

6 alignment length (e.g. including gap characters). Also shown are some percent identities. A percent pairwise alignment identity is deﬁ- ned as (idents / MIN(len1, len2)) where idents is the number of exact identities and len1, len2 are the unaligned lengths of the two COVERAGE sequences. The ”average percent identity”, ”most related pair”, and ”most unrelated pair” of the alignment are the average, maximum,

V and minimum of all (N)(N-1)/2 pairs, respectively. The ”most distant 100% 50% 30% 5% seq” is calculated by ﬁnding the maximum pairwise identity (best relative) for all N sequences, then ﬁnding the minimum of these N numbers (hence, the most outlying sequence). alistat is copyrighted by HHMI/Washington University School of Medicine, 1992-2001, and freely distributed under the GNU General Public License.

V 4.3.2 CE To map ligand binding sites from different source structures, report maker uses the CE program: RELATIVE IMPORTANCE http://cl.sdsc.edu/. Shindyalov IN, Bourne PE (1998) ”Protein structure alignment by incremental combinatorial extension Fig. 9. Coloring scheme used to color residues by their relative importance. (CE) of the optimal path . Protein Engineering 11(9) 739-747. 4.3.3 DSSP In this work a residue is considered solvent accessi- ˚ 2 4 APPENDIX ble if the DSSP program finds it exposed to water by at least 10A , which is roughly the area needed for one water molecule to come in 4.1 File formats the contact with the residue. DSSP is copyrighted by W. Kabsch, C. Files with extension “ranks sorted” are the actual trace results. The Sander and MPI-MF, 1983, 1985, 1988, 1994 1995, CMBI version fields in the table in this file: by [email protected] November 18,2002,

• alignment# number of the position in the alignment http://www.cmbi.kun.nl/gv/dssp/descrip.html. • residue# residue number in the PDB file 4.3.4 HSSP Whenever available, report maker uses HSSP ali- • type amino acid type gnment as a starting point for the analysis (sequences shorter than • rank rank of the position according to older version of ET 75% of the query are taken out, however); R. Schneider, A. de • variability has two subfields: Daruvar, and C. Sander. ”The HSSP database of protein structure- 1. number of different amino acids appearing in in this column sequence alignments.” Nucleic Acids Res., 25:226–230, 1997. of the alignment http://swift.cmbi.kun.nl/swift/hssp/ 2. their type • rho ET score - the smaller this value, the lesser variability of 4.3.5 LaTex The text for this report was processed using LATEX; this position across the branches of the tree (and, presumably, Leslie Lamport, “LaTeX: A Document Preparation System Addison- the greater the importance for the protein) Wesley,” Reading, Mass. (1986). • cvg coverage - percentage of the residues on the structure which 4.3.6 Muscle When making alignments “from scratch”, report have this rho or smaller maker uses Muscle alignment program: Edgar, Robert C. (2004), • gaps percentage of gaps in this column ”MUSCLE: multiple sequence alignment with high accuracy and high throughput.” Nucleic Acids Research 32(5), 1792-97. 4.2 Color schemes used The following color scheme is used in figures with residues colored http://www.drive5.com/muscle/ by cluster size: black is a single-residue cluster; clusters composed of more than one residue colored according to this hierarchy (ordered 4.3.7 Pymol The figures in this report were produced using by descending size): red, blue, yellow, green, purple, azure, tur- Pymol. The scripts can be found in the attachment. Pymol quoise, brown, coral, magenta, LightSalmon, SkyBlue, violet, gold, is an open-source application copyrighted by DeLano Scien- bisque, LightSlateBlue, orchid, RosyBrown, MediumAquamarine, tific LLC (2005). For more information about Pymol see DarkOliveGreen, CornflowerBlue, grey55, burlywood, LimeGreen, http://pymol.sourceforge.net/. (Note for Windows tan, DarkOrange, DeepPink, maroon, BlanchedAlmond. users: the attached package needs to be unzipped for Pymol to read The colors used to distinguish the residues by the estimated the scripts and launch the viewer.) evolutionary pressure they experience can be seen in Fig. 9. 4.4 Note about ET Viewer 4.3 Credits Dan Morgan from the Lichtarge lab has developed a visualization 4.3.1 Alistat alistat reads a multiple sequence alignment from the tool specifically for viewing trace results. If you are interested, please file and shows a number of simple statistics about it. These stati- visit: stics include the format, the number of sequences, the total number of residues, the average and range of the sequence lengths, and the http://mammoth.bcm.tmc.edu/traceview/

7 The viewer is self-unpacking and self-installing. Input files to be used 4.7 Attachments with ETV (extension .etvx) can be found in the attachment to the The following files should accompany this report: main report. • 1hvqA.complex.pdb - coordinates of 1hvqA with all of its interacting partners 4.5 Citing this work • 1hvqA.etvx - ET viewer input file for 1hvqA The method used to rank residues and make predictions in this report • can be found in Mihalek, I., I. Res,ˇ O. Lichtarge. (2004). ”A Family of 1hvqA.cluster report.summary - Cluster report summary for Evolution-Entropy Hybrid Methods for Ranking of Protein Residues 1hvqA by Importance” J. Mol. Bio. 336: 1265-82. For the original version • 1hvqA.ranks - Ranks file in sequence order for 1hvqA of ET see O. Lichtarge, H.Bourne and F. Cohen (1996). ”An Evolu- • 1hvqA.clusters - Cluster descriptions for 1hvqA tionary Trace Method Defines Binding Surfaces Common to Protein • 1hvqA.msf - the multiple sequence alignment used for the chain Families” J. Mol. Bio. 257: 342-358. 1hvqA report maker itself is described in Mihalek I., I. Res and O. • Lichtarge (2006). ”Evolutionary Trace Report Maker: a new type 1hvqA.descr - description of sequences used in 1hvqA msf of service for comparative analysis of proteins.” Bioinformatics • 1hvqA.ranks sorted - full listing of residues and their ranking 22:1656-7. for 1hvqA • 1hvqA.1hvqNAG3.if.pml - Pymol script for Figure 5 4.6 About report maker • 1hvqA.cbcvg - used by other 1hvqA – related pymol scripts report maker was written in 2006 by Ivana Mihalek. The 1D ran- • 1hvqA.1hvqNAG1.if.pml - Pymol script for Figure 6 king visualization program was written by Ivica Res.ˇ report maker • 1hvqA.1hvqNAG2.if.pml - Pymol script for Figure 7 is copyrighted by Lichtarge Lab, Baylor College of Medicine, Houston.