HIV Databases HIV Databases home HIV Databases home
HIV sequence database



Neutralization Index Explanation

Neutralization potency (NP) indicates the virus tier (neutralization index, NI) that an anti-HIV-1 serum (or antibody) can be expected to neutralize. It is defined as the value for the fitted logistic function with equal probability of neutralizing and not neutralizing a virus. Graphically, the NI value is where the fitted curve intersects 50% probability (equal odds of neutralizing and not neutralizing Envs tested).

Objective

The NI tool converts serum or antibody neutralization outcomes to a tiered scale, similar to the virus tiers in common use. For instance, Tier 1 viruses are the mostly readily neutralized, and Tier 3 viruses are the most difficult to neutralize. Neutralization of Tier 2 viruses remains a highly desirable goal.

The Neutralization Index uses logistic regression to compute a tier-like score for anti-HIV-1 antibodies or sera. To do this, it requires that your assay results include viruses that we have already analyzed. To be recognizable, the names of these viruses should match the names we use. Our lists of virus names are linked below.

Resulting NI values indicate the virus tier that each serum or antibody can be expected to neutralize. The p-value associated with each score gives the false-positive probability for rejecting the null hypothesis of no slope in the inferred logistic curve, i.e. no significant difference between Envs neutralized and not neutralized. p-values above some predetermined cutoff (the significance level of the test, e.g., p >0.1 or p >0.2) indicate unreliable.

Virus Scores

To interpret NI analysis results as tier-like scores, we computed the geometric mean ID50 titer for each Env against a large panel, which has already been published. The geometric mean ID50s are rescaled such that the lowest Tier 2 ID50 maps to 2 and the lowest Tier 3 ID50 maps to 3. This is a simple linear transformation, and helps to calibrate results as tiers. For this reason, NI analysis requires that some viruses with tier-like scores available are included in neutralization assays. Better results are obtained by using more of these viruses. The resulting p-values tend to be lower (more significant), because more information powers the regression analysis.

NI analysis also includes two hypothetical Envs. One ("Kleenex") is always neutralized and has a score of 0. The other ("Death Star") is never neutralized and has a score of 5. The two hypothetical extremes are set as constant values for all inputs and not shown in results.

Input

The input for NI computations is a table of neutralization titers, separated (delimited) by either commas, spaces, or tab characters. To specify the input, you may either copy and paste the table into the text box or upload the file using the "Browse..." button.

The input data file should be formatted as described below. For an example, please try the "Sample Input" option.

  • Columns correspond to HIV-1 Envelope-pseudotyped viruses (Envs) used in the assay*.
  • Rows correspond to the samples tested, either serum/plasma or monoclonal antibody.
  • You can indicate censored values using greater than ('>') or less than ('<') symbols, which will be removed before analysis. That is, '<10' will be treated as '10' and '<50' will be treated as '50'.
  • Use 'NA' (without quotation marks) in a cell indicates a missing value.

*To compute the NI values from your sample, at least some of your virus names (data input columns) should match names of viruses that have been calibrated. Links to the lists of our virus names are given below.

Options
  • Plot results

    When this option is enabled, each row in the input data will be used to generate a figure to show neutralization outcomes and the inferred logistic function.

    Each point on the plot shows a virus assayed. The points are colored to suggest virus tier, from grey (easiest to neutralize) through red (hardest to neutralize). The viruses are divided into two groups, those neutralized (bottom) and resistant (top).

    For a given input, each row is used to make one plot. Note that the x-axis values for the viruses will be the same across all plots. Instead, only the neutralization outcomes change. This is because the x-axis values, tier-scaled virus titers, are precalculated.

    The curved blue line shows the inferred logistic function and the dashed line indicates the NI value, which separates the two groups.

  • Sera (Plasma) or Antibodies

    Serum (or plasma) neutralization assays involve dilution series and lower values indicate less neutralization. For Antibodies, units are µg/ml and lower values indicate greater neutralization.

    A sample input data file for use with the Antibodies option is provided as a sample input. Data for that sample are from Kong et al. JVI 89:2659-71, 2015.

  • M Group or Clade C

    Use this option to specify whether the background viruses represent the HIV-1 M group or only (southern African) clade C.

    A sample input data file for use with the Clade C option is provided as a sample input. Data for that sample are from Rademeyer et al. PLoS Path 12:e1005742, 2016.

  • Raw Data Values vs. Binary Transformed

    This option specifies whether you are submitting raw neutralization titers or binary-transformed values. The Raw Data Values option indicates you wish to analyze ID50 or IC50 neutralization titers using the default cutoffs.

    The Binary Transformed alternative indicates that you have applied your own cutoffs to the data and data values consist of only binary outcomes, using "1" and "0" to indicate assay outcomes that resist neutralization or not, respectively. That is, a value of "1" means the virus was not neutralized (i.e. resistant) and a value of "0" means that it was sensitive to neutralization. Missing values can be indicated using "NA".

    A sample input data file for use with the Binary option is provided as a sample input.

Virus Lists

Please note:

  • Virus names in your input data should approximately match the names in our list to ensure they are all used. Our program will attempt to match slightly-different names to ours.
  • NI analysis will provide best results by including as many of the Envs from our list as possible, preferrably 10 or more. Including only very a few of these Envs is unlikely to produce meaningful results, because the analysis is underpowered to detect significant differences between viruses neutralized and not neutralized.
  • It is important to review any mismatched virus names summarized in red, because small differences in the names can exclude an Env that should otherwise be used in the analysis.

M Group Env List

The Env names, together with their clades and computed NI values, are listed: M Group Env List.

Clade C Env List

Clade C Env names, together with their GenBank accession numbers and computed NI values, are listed: Clade C Env List.

External Background Resources

Background information about logistic regression (Wikipedia).

Background information about p-value and statistical inference (Wikipedia).

The reference publication for the M group Envs is Hraber et al. 2014, PMID 25142591. Supporting data are available online from Neutralization Serotype Discovery Panel.

Reference publication for the clade C Envs is Rademeyer et al. 2016, PMID 27434311.

Reference

P Hraber, B Korber, K Wagh, D Montefiori, M Roederer. A single, continuous metric to define tiered serum neutralization potency against HIV. eLife e31805, 2018. doi:10.7554/eLife.31805. PMID 29350181

Tool Interface

Return to Neutralization Index input page.

 

last modified: Tue Jul 17 09:58 2018


Questions or comments? Contact us at seq-info@lanl.gov.

 
Operated by Triad National Security, LLC for the U.S. Department of Energy's National Nuclear Security Administration
© Copyright Triad National Security, LLC. All Rights Reserved | Disclaimer/Privacy

Dept of Health & Human Services Los Alamos National Institutes of Health