Usage

(1) Guidance for submission:

In the "Missense Mutation Pathogenicity Prediction" row, users can input multiple variants once. Notably, there are two format requirements, as below:

(1) Due to the limitation of computing resources, the number of input variants should be less than 100.

(2) the input variants must be in VCF format as follows:

#CHROM POS REF ALT

1 1168196 C T

2 1481120 G T

3 9730718 A G

4 515710 G C

5 218471 A T

6 109787446 C T

7 195679 C T

8 1719221 A G

9 215012 C A

11 532671 G C

12 862859 C G

13 20567666 G C

14 20915433 C T

15 23086278 G C

16 136789 A G

16 138725 T A

16 138755 C T

16 139765 G A

16 139784 G C

17 436083 T C

19 620430 C T

20 398445 C G

20 400365 A G

21 16337059 G A

21 16339874 T C

22 17584454 G A

etc.

OR, users can upload "variants.vcf" file with the correct format.

(3) According to the variants information input by the user, the programs will calculate the characteristics of the variants, including:

Variant-level annotation features using Ensembl VEP v104 with critical pluing dbNSFP V4.1a;

We also applied other databases, such as ClinVar(202012); HGMD-PUBLIC(20204);dbSNP(v154);

Amino Acid-level embeddings. Two tools (ESM-1b and ProtT5-XL-U50) are adopted;

Genome-level annotation features, such as gnomAD_AFR_AF and ExAC_SAS_AF.

(4) Taking about 100 variants as one unit, we listed the time required for each part during the prediction procedure as below:

Ensembl VEP v104 with critical pluing dbNSFP V4.1a (need ~3 minutes);

Gererating protein sequence using API from web (need ~ 4 minutes);

ESM-1b amino acid-level embeddings (need ~ 5 minutes);

ProtT5-XL-U50 amino acid-level embeddings (need ~ 5 minutes);

Missense mutation pathogenicity prediction using ConsMM model (need ~ 2 minutes);

Missense mutation pathogenicity prediction using EvoIndMM model (need ~ 2 minutes).

(5) In the manuscript, for a variant, we labelled "pathogenic" and "neutral" as 1 and 0, respectively, for ConsMM and EvoIndMM model. Accordingly, the final score in the result table refers to the probability of variant belonging to "pathogenic variant" class. As for the phenotype, we take 0.5 as the prediction cutoff threshold value in the manuscript. That is, if the final score is larger than 0.5, the corresponding mutation is predicted as "pathogenic", and others as "neutral".

Accordingly, users shouldn't wait for results on the waiting page. Once the prediction finished, the results link would send to your email address.

In view of the limitation of computing storage resources, the user's prediction results will be kept for 72 hours. Please download and save the result file when the prediction finished.

(2) Feature descriptions:

Using Ensembl VEP v104 with some plugins (such as dbNSFP4 verison 4.1a, dbSNP version 154, COSMIC version 92, etc), we can get multiple annotation information for input variants. Here, we listed some typical items as below:

(1) MetaSVM_score: Support vector machine (SVM) based ensemble prediction score, which incorporated 10 scores (SIFT, PolyPhen-2 HDIV, PolyPhen-2 HVAR, GERP++, MutationTaster, Mutation Assessor, FATHMM, LRT, SiPhy, PhyloP) and the maximum frequency observed in the 1000 genomes populations. Larger value means the SNV is more likely to be damaging. Scores range from -2 to 3 in dbNSFP.

(2) MutPred_Top5features: Top 5 features (molecular mechanisms of disease) as predicted by MutPred with p values. MutPred_score > 0.5 and p < 0.05 are referred to as actionable hypotheses. MutPred_score > 0.75 and p < 0.05 are referred to as confident hypotheses. MutPred_score > 0.75 and p < 0.01 are referred to as very confident hypotheses.

(3) REVEL_score: REVEL is an ensemble score based on 13 individual scores for predicting the pathogenicity of missense variants. Scores range from 0 to 1. The larger the score the more likely the SNP has damaging effect.

(4) FATHMM_converted_rankscore:FATHMMori scores were first converted to FATHMMnew=1-(FATHMMori+16.13)/26.77, then ranked among all FATHMMnew scores in dbNSFP. The rankscore is the ratio of the rank of the score over the total number of FATHMM new scores in dbNSFP. If there are multiple scores, only the most damaging (largest) rankscore is presented. The scores range from 0 to 1.

(5) AF: Frequency of existing variant in 1000 Genomes combined population.

(6) gnomAD_SAS_AF: Frequency of existing variant in gnomAD exomes South Asian population.

(7) ExAC_EAS_AF: Frequency of existing variant in ExAC East Asian population.

(8) ExAC_nonTCGA_AFR_AF:Adjusted Alt allele frequency (DP >= 10 & GQ >= 20) in African & African American ExAC_nonTCGA samples.

(9) gnomAD_exomes_AFR_nhomalt: Count of individuals with homozygous alternative allele in the African/African American gnomAD exome samples v2.1.1.

For other feature descriptions,please see the supplementary material of this manuscript. OR,please see http://grch37.ensembl.org/info/docs/tools/vep/script/vep_plugins.html and https://sites.google.com/site/jpopgen/dbNSFP.

1.Explanations of variants outputs

2.Example outputs