TSARseqNovo

algorithm · Transformer (AR)

TSARseqNovo: algorithm · Transformer (AR). Semi-autoregressive

Semi-autoregressive

Kind algorithm
Deep learning yes
Acquisition DDA
Family Transformer (AR)

Code

Live stars, open issues and last-push figures are on the Code activity chart.

Data

What the papers describing this method trained it on and reported its results on, as each paper states it. A version is named only where the paper names one.

Trained on

Evaluated on

Reported comparison

The comparison table this method’s own papers print, standardised: every value on a 0-1 scale, methods down the side, the measure and then the species across. These are numbers papers report about themselves and their baselines. They are not a leaderboard, and they do not compare across tables: each was produced by a different group, on the dataset named in its corner, with each baseline either retrained, run from released weights or quoted from another paper. Where the paper says which, it follows the method’s name (hover it for the sentence); most papers do not say. Bold is the best value in a column and underline the runner-up, our ranking rather than the paper’s own marks.

Table1

A transformer-based semi-autoregressive framework for high-speed and accurate de novo peptide sequencing, page 5: Comparative Results of TSARseqNovo, CasaNovo, and π-HelixNovo on the Nine-Species Dataset

Nine-species benchmark Peptide precision
Candidatus Thiodiazotropha endoloripes Homo sapiens Apis mellifera Mus musculus Solanum lycopersicum Methanosarcina mazei Bacillus subtilis Saccharomyces cerevisiae Vigna mungo Average
TSARseqNovo 0.419 0.430 0.512 0.532 0.571 0.582 0.612 0.618 0.637 0.535
Casanovo 0.371‡ 0.346‡ 0.408‡ 0.442‡ 0.461‡ 0.486‡ 0.540‡ 0.505‡ 0.523‡ 0.445‡
π-HelixNovo 0.388‡ 0.392‡ 0.473‡ 0.483‡ 0.560‡ 0.560‡ 0.596‡ 0.568‡ 0.623‡ 0.503‡
Nine-species benchmark Amino acid precision
Candidatus Thiodiazotropha endoloripes Homo sapiens Apis mellifera Mus musculus Solanum lycopersicum Methanosarcina mazei Bacillus subtilis Saccharomyces cerevisiae Vigna mungo Average
TSARseqNovo 0.706 0.677 0.755 0.812 0.793 0.783 0.827 0.792 0.835 0.776
Casanovo 0.621‡ 0.585‡ 0.638‡ 0.709‡ 0.674‡ 0.704‡ 0.744‡ 0.705‡ 0.718‡ 0.678‡
π-HelixNovo 0.681‡ 0.665‡ 0.721‡ 0.765‡ 0.771‡ 0.784‡ 0.816‡ 0.768‡ 0.793‡ 0.752‡
Nine-species benchmark Amino acid recall
Candidatus Thiodiazotropha endoloripes Homo sapiens Apis mellifera Mus musculus Solanum lycopersicum Methanosarcina mazei Bacillus subtilis Saccharomyces cerevisiae Vigna mungo Average
TSARseqNovo 0.699 0.683 0.767 0.822 0.786 0.795 0.828 0.799 0.833 0.779
Casanovo 0.624‡ 0.591‡ 0.640‡ 0.709‡ 0.675‡ 0.705‡ 0.745‡ 0.706‡ 0.719‡ 0.679‡
π-HelixNovo 0.668‡ 0.667‡ 0.729‡ 0.767‡ 0.771‡ 0.784‡ 0.817‡ 0.744‡ 0.797‡ 0.749‡

‡ Not printed in the paper: computed from the differences it prints, as described under the table.
The original table prints only TSARseqNovo's scores, each followed by two 'vs' rows giving its improvement in percentage points over CasaNovo and pi-HelixNovo. The CasaNovo and pi-HelixNovo values here are computed as TSARseqNovo minus that improvement; they are not printed in the paper. The table also misspells pi-HelixNovo as 'pi-HelexiNovo'.

Compared against in other papers (1)

1 table from 1 paper that report this method as a baseline, newest first. The caveats of the comparisons above hold twice over: each paper ran, retrained or quoted this method its own way, so its number here is that paper’s claim, not a measurement of the method. Where a paper says which it did, the word follows the method’s name.

RT-GCTnovo: A Peptide De Novo Sequencing Model Incorporating Gated Multi-scale Features and Dynamic Mass Masks (2025) · 1 table

TABLE I

RT-GCTnovo: A Peptide De Novo Sequencing Model Incorporating Gated Multi-scale Features and Dynamic Mass Masks, page 6: Comparison of Average Accuracy (%)

Nine-species benchmark Peptide accuracy Amino acid accuracy
DeepNovo 0.376 0.673
Casanovo 0.449 0.678
TSARseqNovo 0.546 0.775
RT-GCTnovo 0.577 0.798

Paper describing it

Authors (8)

Yang Zhao, Shuo Wang, Jinze Huang, Bo Meng, Dong An, Xiang Fang, Yaoguang Wei, Xinhua Dai

Seen in the charts

Back to the full map

Back to top