{"feature_sets":[{"about":"Physico-chemical properties of a protein sequence (ProtParam-compatible) and its composition","features":[{"name":"length","group":"size","unit":"residues","description":"Number of residues"},{"name":"molecular_weight","group":"size","unit":"Da","description":"Average molecular weight (average residue masses plus one water)","reference":"Gasteiger et al. (2005) Protein identification and analysis tools on the ExPASy server"},{"name":"isoelectric_point","group":"charge","unit":"pH","description":"Theoretical pI: the pH at which the net charge is zero (Bjellqvist pK values with terminal-dependent values)","reference":"Bjellqvist et al. (1993) Electrophoresis 14:1023; (1994) 15:529"},{"name":"net_charge_ph7","group":"charge","unit":"e","description":"Net charge at pH 7.0 with the same pK values","reference":"Henderson–Hasselbalch"},{"name":"charge_density_ph7","group":"charge","unit":"e/residue","description":"Net charge at pH 7.0 per residue"},{"name":"extinction_280_reduced","group":"optics","unit":"M^-1 cm^-1","description":"Molar extinction coefficient at 280 nm with every cysteine reduced: 5500·W + 1490·Y","reference":"Pace et al. (1995) Protein Sci 4:2411"},{"name":"extinction_280_cystines","group":"optics","unit":"M^-1 cm^-1","description":"Molar extinction coefficient at 280 nm with every cysteine pair forming a cystine: + 125 per cystine","reference":"Pace et al. (1995) Protein Sci 4:2411"},{"name":"absorbance_280_1mgml","group":"optics","unit":"AU","description":"Absorbance at 280 nm of a 1 mg/mL solution (1 cm, cysteines reduced)"},{"name":"instability_index","group":"stability","description":"Instability index; above 40 suggests a protein unstable in vitro","reference":"Guruprasad et al. (1990) Protein Eng 4:155"},{"name":"aliphatic_index","group":"stability","description":"Relative volume of aliphatic side chains: X(Ala) + 2.9·X(Val) + 3.9·(X(Ile) + X(Leu)), mole percent","reference":"Ikai (1980) J Biochem 88:1895"},{"name":"gravy","group":"hydropathy","description":"Grand average of hydropathy (Kyte–Doolittle)","reference":"Kyte & Doolittle (1982) J Mol Biol 157:105"},{"name":"aromaticity","group":"hydropathy","unit":"fraction","description":"Fraction of F, W and Y","reference":"Lobry & Gautier (1994) Nucleic Acids Res 22:3174"},{"name":"frac_tiny","group":"classes","unit":"fraction","description":"Fraction of tiny residues (ACGST)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_small","group":"classes","unit":"fraction","description":"Fraction of small residues (ABCDGNPSTV)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_aliphatic","group":"classes","unit":"fraction","description":"Fraction of aliphatic residues (AILV)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_aromatic","group":"classes","unit":"fraction","description":"Fraction of aromatic residues (FHWY)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_nonpolar","group":"classes","unit":"fraction","description":"Fraction of nonpolar residues (ACFGILMPVWY)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_polar","group":"classes","unit":"fraction","description":"Fraction of polar residues (DEHKNQRSTZ)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_charged","group":"classes","unit":"fraction","description":"Fraction of charged residues (BDEHKRZ)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_basic","group":"classes","unit":"fraction","description":"Fraction of basic residues (HKR)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"frac_acidic","group":"classes","unit":"fraction","description":"Fraction of acidic residues (BDEZ)","reference":"EMBOSS pepstats residue classes (Rice et al. 2000)"},{"name":"cysteine_count","group":"composition","unit":"residues","description":"Number of cysteines (disulfide capacity)"},{"name":"entropy_bits","group":"complexity","unit":"bits","description":"Shannon entropy of the amino-acid composition"},{"name":"entropy_normalized","group":"complexity","unit":"fraction","description":"Shannon entropy divided by its maximum, log2(20)"},{"name":"low_complexity_fraction","group":"complexity","unit":"fraction","description":"Fraction of residues SEG masks as low complexity (window 12, 2.2, 2.5, as BLAST)","reference":"Wootton & Federhen (1993) Comput Chem 17:149"},{"name":"ambiguous_fraction","group":"complexity","unit":"fraction","description":"Fraction of letters outside the 20 standard amino acids (X, B, Z, J, U, O)"},{"name":"aa_A","group":"composition","unit":"fraction","description":"Fraction of alanine"},{"name":"aa_C","group":"composition","unit":"fraction","description":"Fraction of cysteine"},{"name":"aa_D","group":"composition","unit":"fraction","description":"Fraction of aspartate"},{"name":"aa_E","group":"composition","unit":"fraction","description":"Fraction of glutamate"},{"name":"aa_F","group":"composition","unit":"fraction","description":"Fraction of phenylalanine"},{"name":"aa_G","group":"composition","unit":"fraction","description":"Fraction of glycine"},{"name":"aa_H","group":"composition","unit":"fraction","description":"Fraction of histidine"},{"name":"aa_I","group":"composition","unit":"fraction","description":"Fraction of isoleucine"},{"name":"aa_K","group":"composition","unit":"fraction","description":"Fraction of lysine"},{"name":"aa_L","group":"composition","unit":"fraction","description":"Fraction of leucine"},{"name":"aa_M","group":"composition","unit":"fraction","description":"Fraction of methionine"},{"name":"aa_N","group":"composition","unit":"fraction","description":"Fraction of asparagine"},{"name":"aa_P","group":"composition","unit":"fraction","description":"Fraction of proline"},{"name":"aa_Q","group":"composition","unit":"fraction","description":"Fraction of glutamine"},{"name":"aa_R","group":"composition","unit":"fraction","description":"Fraction of arginine"},{"name":"aa_S","group":"composition","unit":"fraction","description":"Fraction of serine"},{"name":"aa_T","group":"composition","unit":"fraction","description":"Fraction of threonine"},{"name":"aa_V","group":"composition","unit":"fraction","description":"Fraction of valine"},{"name":"aa_W","group":"composition","unit":"fraction","description":"Fraction of tryptophan"},{"name":"aa_Y","group":"composition","unit":"fraction","description":"Fraction of tyrosine"}],"kind":"protein","name":"gl1x-protein-1","prefix":"p_"}],"gl1f":{"default_scale_q":1000,"features_text":"the GL1F NFT features string: one feature name per line, in input order, as a model NFT stores its inputs","note":"a model trained on a dataset's feature columns takes them in the dataset's column order; quantize new inputs with the model's own scale_q","packing":"one little-endian int32 per input in feature order: the bytes the GL1F runtime's predict functions take","quantization":"q = clamp(round(x · scale_q)) to the int32 range, the GL1F Forest input rule (half rounds away from zero for positive x, toward +∞ as JavaScript's Math.round)"},"molnft_paths":{"0xde3723766bc32dcacd03c17baa400a7b36837eba:2":"/molnft/101D-2"},"record":{"accession":null,"collection_name":"MolNFT v1 / PDB","contract_address":"0xde3723766bc32dcacd03c17baa400a7b36837eba","molecular_id":"101D","molnft_version":"v1","pdb_id":"101D","structure_available":true,"title":"REFINEMENT OF NETROPSIN BOUND TO DNA: BIAS AND FEEDBACK IN ELECTRON DENSITY MAP INTERPRETATION","token_id":"2","url":"/molnft/101D-2"},"sequences":[{"ambiguous":1,"chain_id":"A","chains":["A","B"],"is_primary":true,"length":12,"notes":["12 residues, 1,010.0 Da (1.0 kDa) average mass","Theoretical pI 5.50; net charge -0.3 at pH 7 (near neutral)","Extinction at 280 nm: 125 M⁻¹cm⁻¹ with every cysteine pair a cystine, 0 reduced; A280 of 1 mg/mL ≈ 0.000","Instability index 0.42: classified stable in vitro (below 40)","GRAVY +0.736: hydrophobic on average; aliphatic index 16.7 (higher suggests more thermostable)","SEG masks 100.0% of the residues as low complexity","1 letters outside the 20 standard amino acids are left out of the compositions"],"sequence":"CGCGAATTXGCG","sequence_key":"pdb-chain:A","sequence_keys":["pdb-chain:A","pdb-chain:B"],"sequence_type":"protein","set":"gl1x-protein-1","sha256":"4167f33c47f85cb9744d531a37b77ffb5fa24a0948f29c3c099e1702775cd22a","values":[12,1010.01,5.5049,-0.2696,-0.022467,0,125,0,0.425,16.6667,0.7364,0,0.916667,0.916667,0.166667,0,0.75,0.166667,0,0,0,3,1.93626,0.448008,1,0.083333,0.166667,0.25,0,0,0,0.333333,0,0,0,0,0,0,0,0,0,0,0.166667,0,0,0]}],"sequences_truncated":false,"structure":{"chains":[],"entry_id":"101D","method":"X-RAY DIFFRACTION","predicted":false,"skipped":[{"chain_id":"A","kind":"dna","reason":"structure features describe protein chains"},{"chain_id":"B","kind":"dna","reason":"structure features describe protein chains"}],"status":"complete"}}