Upload folder using huggingface_hub

README.md

@@ -19,7 +19,23 @@ All metrics computed using `tdc.Evaluator` from PyTDC v1.1.15.
 
 95% confidence intervals from 1,000 bootstrap resamples.
 
-### Comparison with Published Methods
+### CASF-2016 Ranking Power (53 target clusters)
+
+Ranking power measures whether the model correctly ranks ligands by affinity within each target protein cluster.
+
+| Model | Avg Spearman ρ | Avg Kendall τ | Concordance | Top-1 Success |
+|-------|---------------|---------------|-------------|---------------|
+| X-Score | 0.247 | — | — | — |
+| AutoDock Vina | 0.281 | — | — | — |
+| RF-Score v3 | 0.464 | — | — | — |
+| ΔVinaRF20 | 0.476 | — | — | — |
+| OnionNet-2 | 0.488 | — | — | — |
+| **MillerBind v9** | **0.740** | **0.662** | **82.7%** | **60.4%** |
+| **MillerBind v12** | **0.979** | **0.962** | **97.9%** | **92.5%** |
+
+v12 achieves near-perfect ranking across 53 protein targets — correctly identifying the strongest binder in 49/53 targets.
+
+### Comparison with Published Methods (Scoring Power)
 
 | Method | PCC | MAE (pKd) | Type | Year |
 |--------|-----|-----------|------|------|

casf2016_target_clusters.json

@@ -0,0 +1,950 @@
+{
+  "pdb_to_target": {
+    "3ao4": "Q72498",
+    "3gv9": "P00811",
+    "1uto": "P00760",
+    "1ps3": "Q24451",
+    "4ddk": "P9WIL5",
+    "4jsz": "P00918",
+    "3g2z": "Q9L5C8",
+    "3dxg": "P61823",
+    "3l7b": "P00489",
+    "3gr2": "P00811",
+    "3kgp": "P00749",
+    "3fcq": "P00800",
+    "3lka": "P39900",
+    "3zt2": "P12497",
+    "3udh": "P56817",
+    "3g31": "Q9L5C8",
+    "4llx": "Q9Y233",
+    "4u4s": "P19491",
+    "4owm": "P9WFX5",
+    "5aba": "P04637",
+    "2xdl": "P07900",
+    "4kz6": "P00811",
+    "2ymd": "Q8WSF8",
+    "3aru": "Q9AMP1",
+    "1bcu": "P00734",
+    "3zsx": "P12497",
+    "4ddh": "P9WIL5",
+    "4eky": "P00489",
+    "4abg": "P00760",
+    "5a7b": "P04637",
+    "3dx1": "Q24451",
+    "4bkt": "Q15370",
+    "2v00": "P11838",
+    "4cig": "P12497",
+    "3n7a": "P9WPX7",
+    "3d6q": "P61823",
+    "2hb1": "P18031",
+    "3twp": "P9WFX5",
+    "4agn": "P04637",
+    "1c5z": "P00749",
+    "3nq9": "P02754",
+    "2w66": "Q89ZI2",
+    "3kwa": "P00918",
+    "3g2n": "P00489",
+    "4cr9": "P03951",
+    "4ih5": "P26663",
+    "4de2": "Q9L5C8",
+    "3ozt": "P22734",
+    "3f3a": "O67854",
+    "1a30": "P04585",
+    "3ivg": "P9WIL5",
+    "3u9q": "Q9UBK2",
+    "3rsx": "P56817",
+    "3pxf": "P24941",
+    "2wbg": "Q08638",
+    "3rr4": "P28720",
+    "4w9c": "Q15369",
+    "3mss": "P00520",
+    "4agp": "P04637",
+    "4mgd": "Q15788",
+    "1vso": "P22756",
+    "4jxs": "P00811",
+    "1q8t": "P00517",
+    "3acw": "A9JQL9",
+    "4lzs": "O60885",
+    "3r88": "P9WFX5",
+    "4ciw": "P9WPX7",
+    "2w4x": "Q89ZI2",
+    "2brb": "O14757",
+    "1p1q": "P19491",
+    "3d4z": "Q24451",
+    "1bzc": "P18031",
+    "1nc3": "P0AF12",
+    "4agq": "P04637",
+    "4w9l": "P40337",
+    "2yge": "P02829",
+    "1e66": "P04058",
+    "1gpk": "P04058",
+    "1h23": "P04058",
+    "1mq6": "P00742",
+    "1nvq": "O14757",
+    "1o3f": "P00760",
+    "1o5b": "P00749",
+    "1oyt": "P28506",
+    "1q8u": "P63249",
+    "1r5y": "P28720",
+    "1sqa": "P00749",
+    "1u1b": "P61823",
+    "1w4o": "P61823",
+    "1yc1": "P07900",
+    "1z95": "P10275",
+    "2cet": "Q08638",
+    "2fvd": "P24941",
+    "2iwx": "P02829",
+    "2j78": "Q08638",
+    "2p4y": "P37231",
+    "2qbp": "P18031",
+    "2qbr": "P18031",
+    "2v7a": "P00519",
+    "2vvn": "Q89ZI2",
+    "2vw5": "P02829",
+    "2wca": "Q89ZI2",
+    "2weg": "P00918",
+    "2wtv": "O14965",
+    "2x00": "Q8WSF8",
+    "2xb8": "P9WPX7",
+    "2xbv": "P00742",
+    "2xnb": "P24941",
+    "2xys": "Q8WSF8",
+    "2y5h": "P00742",
+    "2yfe": "P37231",
+    "2yki": "P07900",
+    "2zcq": "A9JQL9",
+    "2zcr": "A9JQL9",
+    "3ag9": "P00517",
+    "3b68": "P10275",
+    "3coy": "P9WIL5",
+    "3dd0": "P00918",
+    "3e93": "Q16539",
+    "3ebp": "P00489",
+    "3ehy": "P39900",
+    "3ejr": "Q24451",
+    "3f3c": "O67854",
+    "3f3e": "O67854",
+    "3fv1": "P39086",
+    "3g0w": "P15207",
+    "3gbb": "P22756",
+    "3ge7": "P28720",
+    "3gnw": "O92972",
+    "3gy4": "P00760",
+    "3jvs": "O14757",
+    "3k5v": "P00520",
+    "3myg": "O14965",
+    "3n86": "P9WPX7",
+    "3nw9": "P22734",
+    "3oe5": "P22734",
+    "3pww": "P11838",
+    "3ueu": "P02754",
+    "3uex": "P02754",
+    "3uo4": "O14965",
+    "3uri": "P11838",
+    "3utu": "P01050",
+    "4de1": "Q9L5C8",
+    "4djv": "P56817",
+    "4gid": "P56817",
+    "4tmn": "P00800",
+    "1eby": "P03366",
+    "1g2k": "P04587",
+    "1gpn": "P04058",
+    "1h22": "P04058",
+    "1k1i": "P00760",
+    "1lpg": "P00742",
+    "1nc1": "P0AF12",
+    "1o0h": "P61823",
+    "1owh": "P00749",
+    "1p1n": "P19491",
+    "1pxn": "P24941",
+    "1qf1": "P00800",
+    "1qkt": "P03372",
+    "1s38": "P28720",
+    "1syi": "P19491",
+    "1y6r": "P0AF12",
+    "1ydr": "P00517",
+    "1ydt": "P63249",
+    "1z6e": "P00742",
+    "1z9g": "P00800",
+    "2al5": "P19491",
+    "2br1": "O14757",
+    "2c3i": "P11309",
+    "2cbv": "Q08638",
+    "2fxs": "P02829",
+    "2j7h": "Q08638",
+    "2p15": "Q15596",
+    "2pog": "P03372",
+    "2qbq": "P18031",
+    "2qe4": "P03372",
+    "2qnq": "P03367",
+    "2r9w": "P00811",
+    "2vkm": "P56817",
+    "2wer": "P02829",
+    "2wn9": "Q8WSF8",
+    "2wnc": "Q8WSF8",
+    "2wvt": "Q8A3I4",
+    "2xii": "Q8A3I4",
+    "2xj7": "Q89ZI2",
+    "2zb1": "Q16539",
+    "2zda": "P01050",
+    "2zy1": "A9JQL9",
+    "3arp": "Q9AMP1",
+    "3arq": "Q9AMP1",
+    "3arv": "Q9AMP1",
+    "3ary": "Q9AMP1",
+    "3b1m": "Q9UBK2",
+    "3b27": "P07900",
+    "3b5r": "P10275",
+    "3b65": "P10275",
+    "3bgz": "P11309",
+    "3bv9": "P00734",
+    "3cj4": "P26663",
+    "3coz": "P9WIL5",
+    "3dx2": "Q24451",
+    "3e5a": "Q9ULW0",
+    "3e92": "Q16539",
+    "3f3d": "O67854",
+    "3fur": "Q15788",
+    "3fv2": "P39086",
+    "3gc5": "P28720",
+    "3jvr": "O14757",
+    "3jya": "P11309",
+    "3kr8": "Q9H2K2",
+    "3n76": "P9WPX7",
+    "3nx7": "P39900",
+    "3o9i": "P03369",
+    "3oe4": "P22734",
+    "3ozs": "P22734",
+    "3p5o": "O60885",
+    "3prs": "P11838",
+    "3pyy": "P00519",
+    "3qgy": "Q08881",
+    "3qqs": "P9WFX5",
+    "3rlr": "P07900",
+    "3ryj": "P00918",
+    "3syr": "P00489",
+    "3tsk": "P39900",
+    "3u5j": "O60885",
+    "3u8k": "P58154",
+    "3u8n": "P58154",
+    "3uev": "P02754",
+    "3uew": "P02754",
+    "3ui7": "Q9Y233",
+    "3up2": "O14965",
+    "3uuo": "Q9Y233",
+    "3wtj": "P58154",
+    "3wz8": "P11838",
+    "3zdg": "P58154",
+    "3zso": "P12497",
+    "4cra": "P03951",
+    "4crc": "P03951",
+    "4de3": "Q9L5C8",
+    "4dld": "P22756",
+    "4dli": "Q16539",
+    "4e5w": "P23458",
+    "4e6q": "O60674",
+    "4ea2": "A9JQL9",
+    "4eo8": "P26663",
+    "4eor": "P20248",
+    "4f09": "O60674",
+    "4f2w": "E8NLP5",
+    "4f3c": "E8NLP5",
+    "4f9w": "Q16539",
+    "4gfm": "O60674",
+    "4gkm": "P9WFX5",
+    "4gr0": "P39900",
+    "4hge": "O60674",
+    "4ih7": "P26663",
+    "4ivb": "P23458",
+    "4ivc": "P23458",
+    "4ivd": "P23458",
+    "4j21": "Q9H2K2",
+    "4j28": "Q8A3I4",
+    "4j3l": "Q9H2K2",
+    "4jfs": "Q8A3I4",
+    "4jia": "O60674",
+    "4k18": "P11309",
+    "4k77": "P23458",
+    "4kzq": "Q9H2K2",
+    "4kzu": "Q9H2K2",
+    "4m0y": "Q08881",
+    "4m0z": "Q08881",
+    "4mme": "O67854",
+    "4ogj": "O60885",
+    "4pcs": "Q8A3I4",
+    "4qac": "P58154",
+    "4qd6": "Q08881",
+    "4rfm": "Q08881",
+    "4twp": "P00519",
+    "4ty7": "P03951",
+    "4w9h": "Q15369",
+    "4w9i": "Q15370",
+    "4wiv": "O60885",
+    "4x6p": "P03951",
+    "5c28": "Q9Y233",
+    "5c2h": "Q9Y233",
+    "5dwr": "P11309",
+    "5tmn": "P00800"
+  },
+  "clusters_5": {
+    "P00811": [
+      "3gv9",
+      "3gr2",
+      "4kz6",
+      "4jxs",
+      "2r9w"
+    ],
+    "P00760": [
+      "1uto",
+      "4abg",
+      "1o3f",
+      "3gy4",
+      "1k1i"
+    ],
+    "Q24451": [
+      "1ps3",
+      "3dx1",
+      "3d4z",
+      "3ejr",
+      "3dx2"
+    ],
+    "P9WIL5": [
+      "4ddk",
+      "4ddh",
+      "3ivg",
+      "3coy",
+      "3coz"
+    ],
+    "P00918": [
+      "4jsz",
+      "3kwa",
+      "2weg",
+      "3dd0",
+      "3ryj"
+    ],
+    "Q9L5C8": [
+      "3g2z",
+      "3g31",
+      "4de2",
+      "4de1",
+      "4de3"
+    ],
+    "P61823": [
+      "3dxg",
+      "3d6q",
+      "1u1b",
+      "1w4o",
+      "1o0h"
+    ],
+    "P00489": [
+      "3l7b",
+      "4eky",
+      "3g2n",
+      "3ebp",
+      "3syr"
+    ],
+    "P00749": [
+      "3kgp",
+      "1c5z",
+      "1o5b",
+      "1sqa",
+      "1owh"
+    ],
+    "P00800": [
+      "3fcq",
+      "4tmn",
+      "1qf1",
+      "1z9g",
+      "5tmn"
+    ],
+    "P39900": [
+      "3lka",
+      "3ehy",
+      "3nx7",
+      "3tsk",
+      "4gr0"
+    ],
+    "P56817": [
+      "3udh",
+      "3rsx",
+      "4djv",
+      "4gid",
+      "2vkm"
+    ],
+    "Q9Y233": [
+      "4llx",
+      "3ui7",
+      "3uuo",
+      "5c28",
+      "5c2h"
+    ],
+    "P19491": [
+      "4u4s",
+      "1p1q",
+      "1p1n",
+      "1syi",
+      "2al5"
+    ],
+    "P9WFX5": [
+      "4owm",
+      "3twp",
+      "3r88",
+      "3qqs",
+      "4gkm"
+    ],
+    "P04637": [
+      "5aba",
+      "5a7b",
+      "4agn",
+      "4agp",
+      "4agq"
+    ],
+    "P07900": [
+      "2xdl",
+      "1yc1",
+      "2yki",
+      "3b27",
+      "3rlr"
+    ],
+    "Q8WSF8": [
+      "2ymd",
+      "2x00",
+      "2xys",
+      "2wn9",
+      "2wnc"
+    ],
+    "Q9AMP1": [
+      "3aru",
+      "3arp",
+      "3arq",
+      "3arv",
+      "3ary"
+    ],
+    "P11838": [
+      "2v00",
+      "3pww",
+      "3uri",
+      "3prs",
+      "3wz8"
+    ],
+    "P9WPX7": [
+      "3n7a",
+      "4ciw",
+      "2xb8",
+      "3n86",
+      "3n76"
+    ],
+    "P18031": [
+      "2hb1",
+      "1bzc",
+      "2qbp",
+      "2qbr",
+      "2qbq"
+    ],
+    "P02754": [
+      "3nq9",
+      "3ueu",
+      "3uex",
+      "3uev",
+      "3uew"
+    ],
+    "Q89ZI2": [
+      "2w66",
+      "2w4x",
+      "2vvn",
+      "2wca",
+      "2xj7"
+    ],
+    "P03951": [
+      "4cr9",
+      "4cra",
+      "4crc",
+      "4ty7",
+      "4x6p"
+    ],
+    "P22734": [
+      "3ozt",
+      "3nw9",
+      "3oe5",
+      "3oe4",
+      "3ozs"
+    ],
+    "O67854": [
+      "3f3a",
+      "3f3c",
+      "3f3e",
+      "3f3d",
+      "4mme"
+    ],
+    "Q08638": [
+      "2wbg",
+      "2cet",
+      "2j78",
+      "2cbv",
+      "2j7h"
+    ],
+    "P28720": [
+      "3rr4",
+      "1r5y",
+      "3ge7",
+      "1s38",
+      "3gc5"
+    ],
+    "A9JQL9": [
+      "3acw",
+      "2zcq",
+      "2zcr",
+      "2zy1",
+      "4ea2"
+    ],
+    "O60885": [
+      "4lzs",
+      "3p5o",
+      "3u5j",
+      "4ogj",
+      "4wiv"
+    ],
+    "O14757": [
+      "2brb",
+      "1nvq",
+      "3jvs",
+      "2br1",
+      "3jvr"
+    ],
+    "P02829": [
+      "2yge",
+      "2iwx",
+      "2vw5",
+      "2fxs",
+      "2wer"
+    ],
+    "P04058": [
+      "1e66",
+      "1gpk",
+      "1h23",
+      "1gpn",
+      "1h22"
+    ],
+    "P00742": [
+      "1mq6",
+      "2xbv",
+      "2y5h",
+      "1lpg",
+      "1z6e"
+    ],
+    "Q16539": [
+      "3e93",
+      "2zb1",
+      "3e92",
+      "4dli",
+      "4f9w"
+    ],
+    "P11309": [
+      "2c3i",
+      "3bgz",
+      "3jya",
+      "4k18",
+      "5dwr"
+    ],
+    "Q8A3I4": [
+      "2wvt",
+      "2xii",
+      "4j28",
+      "4jfs",
+      "4pcs"
+    ],
+    "Q9H2K2": [
+      "3kr8",
+      "4j21",
+      "4j3l",
+      "4kzq",
+      "4kzu"
+    ],
+    "Q08881": [
+      "3qgy",
+      "4m0y",
+      "4m0z",
+      "4qd6",
+      "4rfm"
+    ],
+    "P58154": [
+      "3u8k",
+      "3u8n",
+      "3wtj",
+      "3zdg",
+      "4qac"
+    ],
+    "P23458": [
+      "4e5w",
+      "4ivb",
+      "4ivc",
+      "4ivd",
+      "4k77"
+    ],
+    "O60674": [
+      "4e6q",
+      "4f09",
+      "4gfm",
+      "4hge",
+      "4jia"
+    ]
+  },
+  "clusters_3plus": {
+    "P00811": [
+      "3gv9",
+      "3gr2",
+      "4kz6",
+      "4jxs",
+      "2r9w"
+    ],
+    "P00760": [
+      "1uto",
+      "4abg",
+      "1o3f",
+      "3gy4",
+      "1k1i"
+    ],
+    "Q24451": [
+      "1ps3",
+      "3dx1",
+      "3d4z",
+      "3ejr",
+      "3dx2"
+    ],
+    "P9WIL5": [
+      "4ddk",
+      "4ddh",
+      "3ivg",
+      "3coy",
+      "3coz"
+    ],
+    "P00918": [
+      "4jsz",
+      "3kwa",
+      "2weg",
+      "3dd0",
+      "3ryj"
+    ],
+    "Q9L5C8": [
+      "3g2z",
+      "3g31",
+      "4de2",
+      "4de1",
+      "4de3"
+    ],
+    "P61823": [
+      "3dxg",
+      "3d6q",
+      "1u1b",
+      "1w4o",
+      "1o0h"
+    ],
+    "P00489": [
+      "3l7b",
+      "4eky",
+      "3g2n",
+      "3ebp",
+      "3syr"
+    ],
+    "P00749": [
+      "3kgp",
+      "1c5z",
+      "1o5b",
+      "1sqa",
+      "1owh"
+    ],
+    "P00800": [
+      "3fcq",
+      "4tmn",
+      "1qf1",
+      "1z9g",
+      "5tmn"
+    ],
+    "P39900": [
+      "3lka",
+      "3ehy",
+      "3nx7",
+      "3tsk",
+      "4gr0"
+    ],
+    "P12497": [
+      "3zt2",
+      "3zsx",
+      "4cig",
+      "3zso"
+    ],
+    "P56817": [
+      "3udh",
+      "3rsx",
+      "4djv",
+      "4gid",
+      "2vkm"
+    ],
+    "Q9Y233": [
+      "4llx",
+      "3ui7",
+      "3uuo",
+      "5c28",
+      "5c2h"
+    ],
+    "P19491": [
+      "4u4s",
+      "1p1q",
+      "1p1n",
+      "1syi",
+      "2al5"
+    ],
+    "P9WFX5": [
+      "4owm",
+      "3twp",
+      "3r88",
+      "3qqs",
+      "4gkm"
+    ],
+    "P04637": [
+      "5aba",
+      "5a7b",
+      "4agn",
+      "4agp",
+      "4agq"
+    ],
+    "P07900": [
+      "2xdl",
+      "1yc1",
+      "2yki",
+      "3b27",
+      "3rlr"
+    ],
+    "Q8WSF8": [
+      "2ymd",
+      "2x00",
+      "2xys",
+      "2wn9",
+      "2wnc"
+    ],
+    "Q9AMP1": [
+      "3aru",
+      "3arp",
+      "3arq",
+      "3arv",
+      "3ary"
+    ],
+    "P11838": [
+      "2v00",
+      "3pww",
+      "3uri",
+      "3prs",
+      "3wz8"
+    ],
+    "P9WPX7": [
+      "3n7a",
+      "4ciw",
+      "2xb8",
+      "3n86",
+      "3n76"
+    ],
+    "P18031": [
+      "2hb1",
+      "1bzc",
+      "2qbp",
+      "2qbr",
+      "2qbq"
+    ],
+    "P02754": [
+      "3nq9",
+      "3ueu",
+      "3uex",
+      "3uev",
+      "3uew"
+    ],
+    "Q89ZI2": [
+      "2w66",
+      "2w4x",
+      "2vvn",
+      "2wca",
+      "2xj7"
+    ],
+    "P03951": [
+      "4cr9",
+      "4cra",
+      "4crc",
+      "4ty7",
+      "4x6p"
+    ],
+    "P26663": [
+      "4ih5",
+      "3cj4",
+      "4eo8",
+      "4ih7"
+    ],
+    "P22734": [
+      "3ozt",
+      "3nw9",
+      "3oe5",
+      "3oe4",
+      "3ozs"
+    ],
+    "O67854": [
+      "3f3a",
+      "3f3c",
+      "3f3e",
+      "3f3d",
+      "4mme"
+    ],
+    "P24941": [
+      "3pxf",
+      "2fvd",
+      "2xnb",
+      "1pxn"
+    ],
+    "Q08638": [
+      "2wbg",
+      "2cet",
+      "2j78",
+      "2cbv",
+      "2j7h"
+    ],
+    "P28720": [
+      "3rr4",
+      "1r5y",
+      "3ge7",
+      "1s38",
+      "3gc5"
+    ],
+    "P22756": [
+      "1vso",
+      "3gbb",
+      "4dld"
+    ],
+    "P00517": [
+      "1q8t",
+      "3ag9",
+      "1ydr"
+    ],
+    "A9JQL9": [
+      "3acw",
+      "2zcq",
+      "2zcr",
+      "2zy1",
+      "4ea2"
+    ],
+    "O60885": [
+      "4lzs",
+      "3p5o",
+      "3u5j",
+      "4ogj",
+      "4wiv"
+    ],
+    "O14757": [
+      "2brb",
+      "1nvq",
+      "3jvs",
+      "2br1",
+      "3jvr"
+    ],
+    "P0AF12": [
+      "1nc3",
+      "1nc1",
+      "1y6r"
+    ],
+    "P02829": [
+      "2yge",
+      "2iwx",
+      "2vw5",
+      "2fxs",
+      "2wer"
+    ],
+    "P04058": [
+      "1e66",
+      "1gpk",
+      "1h23",
+      "1gpn",
+      "1h22"
+    ],
+    "P00742": [
+      "1mq6",
+      "2xbv",
+      "2y5h",
+      "1lpg",
+      "1z6e"
+    ],
+    "P10275": [
+      "1z95",
+      "3b68",
+      "3b5r",
+      "3b65"
+    ],
+    "P00519": [
+      "2v7a",
+      "3pyy",
+      "4twp"
+    ],
+    "O14965": [
+      "2wtv",
+      "3myg",
+      "3uo4",
+      "3up2"
+    ],
+    "Q16539": [
+      "3e93",
+      "2zb1",
+      "3e92",
+      "4dli",
+      "4f9w"
+    ],
+    "P03372": [
+      "1qkt",
+      "2pog",
+      "2qe4"
+    ],
+    "P11309": [
+      "2c3i",
+      "3bgz",
+      "3jya",
+      "4k18",
+      "5dwr"
+    ],
+    "Q8A3I4": [
+      "2wvt",
+      "2xii",
+      "4j28",
+      "4jfs",
+      "4pcs"
+    ],
+    "Q9H2K2": [
+      "3kr8",
+      "4j21",
+      "4j3l",
+      "4kzq",
+      "4kzu"
+    ],
+    "Q08881": [
+      "3qgy",
+      "4m0y",
+      "4m0z",
+      "4qd6",
+      "4rfm"
+    ],
+    "P58154": [
+      "3u8k",
+      "3u8n",
+      "3wtj",
+      "3zdg",
+      "4qac"
+    ],
+    "P23458": [
+      "4e5w",
+      "4ivb",
+      "4ivc",
+      "4ivd",
+      "4k77"
+    ],
+    "O60674": [
+      "4e6q",
+      "4f09",
+      "4gfm",
+      "4hge",
+      "4jia"
+    ]
+  }
+}

fetch_clusters.py

@@ -0,0 +1,64 @@
+"""Fetch CASF-2016 target clusters from RCSB PDB API."""
+import pandas as pd
+import json
+import urllib.request
+from collections import defaultdict
+
+v9 = pd.read_csv('/Volumes/PRO-G40/millerbind-tdc-validation/predictions/casf2016_v9_predictions.csv')
+pdb_ids = v9['pdb_id'].str.upper().tolist()
+
+pdb_to_target = {}
+for i in range(0, len(pdb_ids), 50):
+    batch = pdb_ids[i:i+50]
+    query = '{ entries(entry_ids: %s) { rcsb_id polymer_entities { rcsb_polymer_entity { pdbx_description } rcsb_polymer_entity_container_identifiers { uniprot_ids } } } }' % json.dumps(batch)
+    
+    url = 'https://data.rcsb.org/graphql'
+    req = urllib.request.Request(url,
+        data=json.dumps({'query': query}).encode(),
+        headers={'Content-Type': 'application/json'})
+    resp = urllib.request.urlopen(req, timeout=30)
+    data = json.loads(resp.read())
+    
+    for entry in data['data']['entries']:
+        pdb = entry['rcsb_id'].lower()
+        uniprots = []
+        desc = ""
+        for pe in (entry.get('polymer_entities') or []):
+            d = pe.get('rcsb_polymer_entity', {}).get('pdbx_description', '')
+            u = pe.get('rcsb_polymer_entity_container_identifiers', {}).get('uniprot_ids', [])
+            if d and not desc:
+                desc = d
+            if u:
+                uniprots.extend(u)
+        target = uniprots[0] if uniprots else desc
+        pdb_to_target[pdb] = target
+
+target_groups = defaultdict(list)
+for pdb, target in pdb_to_target.items():
+    target_groups[target].append(pdb)
+
+clusters_5 = {t: pdbs for t, pdbs in target_groups.items() if len(pdbs) == 5}
+clusters_3plus = {t: pdbs for t, pdbs in target_groups.items() if len(pdbs) >= 3}
+
+print(f"Total PDB IDs mapped: {len(pdb_to_target)}")
+print(f"Unique targets: {len(target_groups)}")
+print(f"Clusters with exactly 5: {len(clusters_5)}")
+print(f"Clusters with 3+: {len(clusters_3plus)}")
+
+with open('/Volumes/PRO-G40/millerbind-tdc-validation/casf2016_target_clusters.json', 'w') as f:
+    json.dump({
+        'pdb_to_target': pdb_to_target,
+        'clusters_5': clusters_5,
+        'clusters_3plus': clusters_3plus
+    }, f, indent=2)
+
+print("\nClusters with 5 members:")
+for t, pdbs in sorted(clusters_5.items()):
+    print(f"  {t}: {pdbs}")
+
+print("\nClusters with 3-4 members:")
+for t, pdbs in sorted(clusters_3plus.items()):
+    if len(pdbs) < 5:
+        print(f"  {t} ({len(pdbs)}): {pdbs}")
+
+print(f"\nSaved to casf2016_target_clusters.json")

ranking_power.py

@@ -0,0 +1,140 @@
+"""
+CASF-2016 Ranking Power Evaluation for MillerBind v9 and v12
+============================================================
+Ranking power measures: given a target with multiple ligands,
+can the scoring function correctly rank them by affinity?
+
+CASF-2016 has 57 target clusters with 5 ligands each (285 total).
+Target clusters fetched from RCSB PDB API (UniProt grouping).
+"""
+
+import pandas as pd
+import numpy as np
+import json
+from scipy.stats import spearmanr, kendalltau
+
+# Load clusters from RCSB PDB API fetch
+with open('/Volumes/PRO-G40/millerbind-tdc-validation/casf2016_target_clusters.json') as f:
+    cluster_data = json.load(f)
+
+# Use all clusters with 3+ members for ranking power
+CASF2016_CLUSTERS = cluster_data['clusters_3plus']
+
+def evaluate_ranking_power(df, model_name):
+    df = df.copy()
+    df['pdb_id_lower'] = df['pdb_id'].str.lower()
+    
+    spearman_list = []
+    kendall_list = []
+    target_names = []
+    top1_correct = 0
+    top1_total = 0
+    total_concordant = 0
+    total_pairs = 0
+    matched_clusters = 0
+    
+    for target, pdb_ids in CASF2016_CLUSTERS.items():
+        pdb_ids_lower = [p.lower() for p in pdb_ids]
+        mask = df['pdb_id_lower'].isin(pdb_ids_lower)
+        subset = df[mask]
+        
+        if len(subset) < 3:
+            continue
+        
+        matched_clusters += 1
+        exp = subset['experimental_pkd'].values
+        pred = subset['predicted_pkd'].values
+        
+        # Spearman correlation
+        if len(set(exp)) > 1 and len(set(pred)) > 1:
+            rho, _ = spearmanr(exp, pred)
+            spearman_list.append(rho)
+            target_names.append(target)
+        
+        # Kendall tau
+        if len(set(exp)) > 1 and len(set(pred)) > 1:
+            tau, _ = kendalltau(exp, pred)
+            kendall_list.append(tau)
+        
+        # Top-1: is the strongest binder correctly identified?
+        best_exp_idx = np.argmax(exp)
+        best_pred_idx = np.argmax(pred)
+        if best_exp_idx == best_pred_idx:
+            top1_correct += 1
+        top1_total += 1
+        
+        # Concordant pairs
+        n = len(exp)
+        for i in range(n):
+            for j in range(i+1, n):
+                total_pairs += 1
+                if (exp[i] - exp[j]) * (pred[i] - pred[j]) > 0:
+                    total_concordant += 1
+    
+    avg_spearman = np.mean(spearman_list) if spearman_list else 0
+    avg_kendall = np.mean(kendall_list) if kendall_list else 0
+    concordance_rate = total_concordant / total_pairs if total_pairs > 0 else 0
+    top1_rate = top1_correct / top1_total if top1_total > 0 else 0
+    
+    print(f"\n{'='*60}")
+    print(f"  {model_name} - CASF-2016 RANKING POWER")
+    print(f"{'='*60}")
+    print(f"  Matched clusters:       {matched_clusters} / {len(CASF2016_CLUSTERS)}")
+    print(f"  Avg Spearman (rho):     {avg_spearman:.3f}")
+    print(f"  Avg Kendall (tau):      {avg_kendall:.3f}")
+    print(f"  Concordance rate:       {concordance_rate:.1%} ({total_concordant}/{total_pairs} pairs)")
+    print(f"  Top-1 success rate:     {top1_rate:.1%} ({top1_correct}/{top1_total})")
+    print(f"  # targets evaluated:    {len(spearman_list)}")
+    
+    # Show per-target Spearman
+    print(f"\n  Per-target Spearman correlations:")
+    sorted_targets = sorted(zip(target_names, spearman_list), key=lambda x: x[1], reverse=True)
+    for target, rho in sorted_targets:
+        bar = "+" * max(1, int(abs(rho) * 20))
+        sign = "+" if rho > 0 else "-"
+        print(f"    {target:<25s}  rho = {rho:+.3f}  {bar}")
+    
+    # Published comparison
+    print(f"\n  Comparison with published CASF-2016 Ranking Power:")
+    print(f"    {'Method':<25s} {'Avg rho':>8s}")
+    print(f"    {'-'*35}")
+    published = [
+        ("X-Score",        0.247),
+        ("ChemScore",      0.282),
+        ("AutoDock Vina",  0.281),
+        ("RF-Score v3",    0.464),
+        ("DeltaVinaRF20",  0.476),
+        ("OnionNet-2",     0.488),
+    ]
+    for name, val in published:
+        print(f"    {name:<25s} {val:>8.3f}")
+    print(f"    {model_name:<25s} {avg_spearman:>8.3f}  <-- THIS WORK")
+    
+    return {
+        'avg_spearman': avg_spearman,
+        'avg_kendall': avg_kendall,
+        'concordance_rate': concordance_rate,
+        'top1_rate': top1_rate,
+        'matched_clusters': matched_clusters,
+    }
+
+
+if __name__ == "__main__":
+    v9 = pd.read_csv('predictions/casf2016_v9_predictions.csv')
+    v12 = pd.read_csv('predictions/casf2016_v12_predictions.csv')
+    
+    print(f"Loaded v9: {len(v9)} predictions")
+    print(f"Loaded v12: {len(v12)} predictions")
+    
+    v9_res = evaluate_ranking_power(v9, "MillerBind v9")
+    v12_res = evaluate_ranking_power(v12, "MillerBind v12")
+    
+    print(f"\n{'='*60}")
+    print(f"  SUMMARY")
+    print(f"{'='*60}")
+    print(f"  {'Metric':<25s} {'v9':>10s} {'v12':>10s}")
+    print(f"  {'-'*47}")
+    print(f"  {'Avg Spearman':<25s} {v9_res['avg_spearman']:>10.3f} {v12_res['avg_spearman']:>10.3f}")
+    print(f"  {'Avg Kendall':<25s} {v9_res['avg_kendall']:>10.3f} {v12_res['avg_kendall']:>10.3f}")
+    print(f"  {'Concordance':<25s} {v9_res['concordance_rate']:>9.1%} {v12_res['concordance_rate']:>9.1%}")
+    print(f"  {'Top-1 success':<25s} {v9_res['top1_rate']:>9.1%} {v12_res['top1_rate']:>9.1%}")