IPC 2.0 (isoelectric point calculator)

What is IPC 2.0?

The isoelectric point (pI) is the pH where a protein or peptide carries zero net charge. It determines how a molecule behaves during isoelectric focusing, ion exchange chromatography, and 2D gel electrophoresis. Predicting pI from sequence requires knowing the pKa of every ionizable group, which varies depending on the pKa scale used.

IPC 2.0 (Isoelectric Point Calculator 2.0) by Kozlowski combines 19 pKa scales with machine learning to predict pI more accurately than any single Henderson-Hasselbalch calculation. The SVR models trained on 2,324 proteins and 119,092 peptides achieve RMSD values of 0.85 and 0.23 respectively. A SepConv2D deep learning model pushes peptide accuracy further (RMSD 0.22). Version 2.0 also introduced per-residue pKa prediction, estimating the dissociation constant of each ionizable residue in context using an MLP-SVR ensemble.

When to use IPC 2.0 vs simpler calculators

For a quick fixed-model estimate, the pI Calculator runs in the browser using fixed EMBOSS pKa values. Protein Parameters provides a broader sequence property report. IPC 2.0 is the better choice when:

• Accuracy matters (SVR and DL models outperform any single pKa scale)
• Comparing predictions across multiple scales helps assess confidence
• Per-residue pKa values are needed for understanding charge distribution
• Batch processing hundreds of sequences at once

How to use IPC 2.0 online

IPC 2.0 on ProteinIQ predicts the isoelectric point of proteins and peptides from amino acid sequence. Paste sequences in FASTA format (or fetch them from UniProt), choose a prediction method, and get pI values across up to 19 pKa scales plus machine learning predictions. The residue pKa predictor reports D, E, H, K, Y, and terminal sites; it does not predict cysteine or arginine side-chain pKa.

Pricing

A run costs 1 credit for the selected predictor. Predictor and scale selection do not change this price. The exact quote appears before submission.

Inputs

Input	Description
Protein/Peptide Sequences	One or more amino acid sequences in FASTA format. Also accepts .fasta, .faa, .fa, .fas, .txt, or native .csv datasets. UniProt IDs can be fetched directly using the batch fetcher.

FASTA preserves full descriptions and ignores lines beginning with #. FASTA and CSV sequences require uppercase amino acid letters without stop symbols or gaps. Native FASTA and CSV readers substitute B with N, Z with Q, and U with C. Plain sequence text accepts lowercase letters through its native parser.

For pI CSV, use two columns: exp_pI,sequence. The first column contains an experimental pI value for each sequence. Residue pKa CSV uses the native 12-column dataset layout: pKa,uncertinity,pdb,chain,aa,position,monomer,trimer,pentamer,heptamer,nonamer,sequence. Select IPC2_pKa for this dataset. Deep learning CSV runs require at least 15 rows. A pI CSV run returns pI results; residue pKa requires its own dataset or sequence input.

The file limit is 50 MiB. Free accounts can submit up to 25 sequences and Lite accounts up to 1,000; the workspace applies the current plan limits before submission.

Settings

Parameter	Type	Default	Description
Predictor (method)	enum	IPC1_ALL	Select a native pI predictor, all pI predictors, residue pKa only, or the IPC1 report.
IPC1 pKa scale (scale)	enum	ALL	Select any of the 19 listed scales or all scales. This controls the IPC1 report; trained models retain their fixed native feature sets.

Predictor options

Predictor	Best for	What it does
IPC1_ALL	General use	Henderson-Hasselbalch with all 19 pKa scales (18 literature scales + ProMoST). Fastest option.
IPC2_SVR_protein	Proteins	SVR trained on 2,324 proteins from SWISS-2DPAGE and PIP-DB. Reports all 19 scales plus the SVR prediction.
IPC2_SVR_peptide	Peptides	SVR trained on 119,092 peptides from HiRIEF experiments. Reports all 19 scales plus the SVR prediction.
IPC2_DL_peptide	Peptides	SepConv2D deep learning model. Sequences above 60 residues undergo native truncation; the report retains the sequence used for prediction.
ALL	Method comparison	Runs IPC1, both SVR predictors, and peptide deep learning. Sequence inputs also receive residue pKa predictions.
IPC2_pKa	Residue analysis	Runs the 9-MLP plus SVR residue pKa predictor without pI prediction. Supports the native residue pKa CSV dataset.
IPC1_REPORT	Classical report	Runs the original IPC1 FASTA report with molecular weight and native average pI. Includes a 2D-PAGE PNG when the native density calculation can generate it.

Results

The Results tab shows a spreadsheet with one row per sequence:

Column	Description
ID	Full sequence description from the FASTA header, or the native plain-input description.
Length	Number of amino acid residues.
MW	Molecular weight in Daltons.
pI_[scale]	Predicted pI for each pKa scale (e.g., pI_IPC2_protein, pI_Bjellqvist, pI_ProMoST).
pI_SVR_protein	SVR prediction optimized for proteins (when SVR or ALL method selected).
pI_SVR_peptide	SVR prediction optimized for peptides (when SVR or ALL method selected).
pI_DL_peptide	Deep learning prediction (when DL or ALL method selected).
Avg_pI	Average pI across all calculated scales (excluding Patrickios).

The Per-Residue pKa tab shows the sites reported by the native residue predictor. For FASTA with D, E, H, K, or Y at either end, native alanine padding changes the report sequence and position numbering. These positions are retained without conversion to original sequence positions:

Column	Description
Sequence	Which input sequence this residue belongs to.
Position	Native 1-based position in the reported sequence, including terminal alanine padding. The native site and sequence are retained as additional columns.
Residue	Amino acid type (D, E, H, K, Y, or the terminal residue).
pKa	Predicted dissociation constant from the MLP-SVR ensemble.

Downloadable outputs

Native predictor reports are available as FASTA, CSV, or residue pseudo-FASTA, alongside ipc2_results.json and run.log. The pseudo-FASTA report has three lines per record: description, native sequence, and semicolon-separated residuePosition_pKa sites. IPC1 reports retain three decimal places; SVR and residue pKa reports retain five. Deep learning follows the precision written by its native report.

The JSON file contains the spreadsheet projection. For comparisons requiring the complete native result, use the predictor report. The IPC1_REPORT plot is conditional on the native density calculation; small or degenerate datasets may produce no PNG.

Interpreting pI values

Most proteins have pI between 4 and 10. Knowing where a protein falls relative to pH 7 determines its behavior in common buffers:

pI range	Character	At pH 7.4 (physiological)
<5	Strongly acidic	Net negative charge
5 to 7	Weakly acidic	Slight negative charge
7 to 9	Weakly basic	Slight positive charge
>9	Strongly basic	Net positive charge

Differences between pKa scales show sensitivity to the chosen constants. Agreement is not a calibrated confidence score. The protein and peptide SVR models were trained on different experimental datasets, so choose the model appropriate to the sequence and validate consequential conclusions experimentally.

Deep learning can produce values outside the usual pI range for long or highly charged sequences. ProteinIQ preserves the reported values. For proteins, compare the protein SVR result rather than treating a truncated peptide-model prediction as a reliable pI estimate.

Interpreting per-residue pKa

Per-residue pKa values show how sequence context shifts each ionizable group's dissociation constant away from its "standard" textbook value. For example, an aspartate surrounded by other negative residues will have a higher pKa (harder to deprotonate), while one near positive charges will have a lower pKa.

These values are useful for identifying residues with unusual protonation behavior, which matters for enzyme active sites, pH-dependent conformational changes, and designing mutations that shift pI.

How does IPC 2.0 work?

Henderson-Hasselbalch method

The classical approach calculates net protein charge as a function of pH using:

$\text{pH} = \text{p}K_a + \log\frac{[\text{A}^-]}{[\text{HA}]}$

Starting at pH 6.51, a bisection algorithm adjusts pH up or down until the sum of positive charges (Lys, Arg, His, N-terminus) equals negative charges (Asp, Glu, Cys, Tyr, C-terminus) within a precision of 0.01 pH units.

Each pKa scale assigns different dissociation constants to these groups, producing different pI estimates. The Bjellqvist scale additionally uses position-specific pKa values for N-terminal and C-terminal residues (e.g., N-terminal Ala has pKa 7.59 instead of the generic 7.5), which matters for proteins starting with A, M, S, P, T, V, or E.

ProMoST uses a separate model with three pKa values per residue (N-terminal, middle, C-terminal positions), giving it slightly different behavior from the other 18 scales.

pKa scales

IPC 2.0 includes 19 pKa scales:

Scale	Origin
IPC2_protein	Optimized for proteins (Kozlowski 2021)
IPC2_peptide	Optimized for peptides (Kozlowski 2021)
IPC_protein	Original IPC scale (Kozlowski 2016)
IPC_peptide	Original IPC scale (Kozlowski 2016)
ProMoST	Position-specific model (Halligan 2004)
Bjellqvist	2D electrophoresis standard with terminal corrections
EMBOSS	EMBOSS software suite
DTASelect	Proteomics analysis
Grimsley	Experimental NMR measurements
Lehninger	Biochemistry textbook
Solomon	Protein chemistry
Sillero	Theoretical calculations
Rodwell	Biochemistry reference
Thurlkill	NMR measurements in pentapeptides
Toseland	Statistical analysis of PDB structures
Nozaki	Model compound measurements
Dawson	Data compilation
Wikipedia	General reference values
Patrickios	Simplified (only D/E/K/R, ignores C/H/Y)

The Avg_pI column averages all scales except Patrickios, whose simplified model (ignoring cysteine, histidine, and tyrosine) makes it an outlier.

Support vector regression

The SVR models take a 19-dimensional feature vector (pI predictions from all 18 H-H scales + ProMoST) and feed it into a support vector machine with RBF kernel. The protein SVR was trained on 2,324 proteins from SWISS-2DPAGE and PIP-DB. The peptide SVR was trained on 119,092 peptides from high-resolution isoelectric focusing (HiRIEF) experiments.

Because the SVR sees all 19 scale predictions simultaneously, it learns which scales are most informative for different sequence compositions. It consistently outperforms any individual scale.

SepConv2D deep learning

The deep learning model uses a separable convolutional architecture with four input channels:

• One-hot encoding: 22 x 60 matrix of amino acid identity
• AAindex features: 15 physicochemical indices per position plus hydrophobicity
• Composition vector: Amino acid counts repeated across sequence length
• IPC predictions: pI from all scales + SVR, giving the model calibrated priors

Sequences longer than 60 residues undergo truncation that preferentially removes non-charged residues (Ala, Gly, Leu, etc.) while preserving ionizable residues that determine pI. This truncation uses random selection among non-charged positions, so predictions for long sequences may vary slightly between runs.

Per-residue pKa (MLP-SVR ensemble)

The per-residue pKa predictor estimates the dissociation constant of each ionizable residue (D, E, H, K, Y) plus the N- and C-terminal groups. It uses a stacking ensemble of 9 multilayer perceptrons (MLPs) feeding into a final SVR:

• 6 sequence-only MLPs operating on 5-mers through 15-mers around the target residue
• 3 AAindex-augmented MLPs operating on 3-mers through 7-mers

Each MLP encodes the local sequence context around an ionizable residue using one-hot encoding (and optionally AAindex physicochemical features). The 9 MLP predictions are stacked as features for the final SVR, which outputs a single pKa value. Charged N- and C-terminal residues are flanked with alanine padding before prediction to ensure they are treated as terminal groups.