Bioinformatics & multi-omics

Bioinformatics and multi-omics turn biological samples into interpreted data: sequence a genome, profile single cells, map molecules in tissue, then resolve the result with computation. The throughput is staggering — Illumina’s NovaSeq X roadmap lifts maximum output from 25 to 35 billion reads, a 40% gain, and completes 14 billion reads in 20–22 hours, about 30% faster on whole-genome workflows. The frontier is no longer reading DNA but interpreting it across layers: single-cell and spatial transcriptomics, proteomics and metabolomics, fused by foundation models. Clinical multimodal platforms now reason over millions of records — Tempus AI’s Lens runs analyses against more than 8.5 million de-identified patient records — while Chinese instrument makers have closed the hardware gap, MGI Tech holding roughly 70% of China’s sequencer market.

Key directions of bioinformatics & multi-omics:

1. High-throughput sequencing and analysis (Sequencing & Analysis): short- and long-read platforms with integrated variant-calling pipelines like DRAGEN.
2. Single-cell omics (Single-Cell Omics): profiling expression one cell at a time to resolve heterogeneity invisible to bulk assays.
3. Spatial multi-omics (Spatial Omics): mapping transcripts and proteins in their tissue location with Xenium, Visium and Stereo-seq.
4. Biological AI and bioinformatics services (Biological AI): foundation models and CRO-scale analysis turning multimodal data into prioritized targets.

Sectoral value chain#

[sample] ──> [library prep] ──> [sequencing] ──> [primary analysis]
                                  │
                          (reads + QC)
                                  │
                                  ▼
[interpreted report] <─── [multi-omics integration] <─────┘

Value chain levels#

Cross-cutting technologies of the sector:

• Cloud bioinformatics pipelines (Cloud Pipelines): scalable DRAGEN-style analysis across cohorts.
• Foundation models for biology (Foundation Models): sequence and multimodal models for prediction.
• Adaptive sampling and basecalling (Adaptive Sampling): real-time target enrichment on long-read flow cells.

The stack spans sequencing chemistry, single-cell and spatial capture, and AI interpretation.

1. High-throughput sequencing (NovaSeq & DNBSEQ):
   • Illumina’s NovaSeq X reaches 35 billion reads with a Q70 quality score.
   • MGI Tech’s DNBSEQ instruments hold roughly 70% of China’s market, with the E25 cleared by NMPA.
2. Single-cell and spatial omics (Chromium & Xenium):
   • 10x Genomics Chromium resolves expression cell by cell; Xenium maps transcripts in tissue.
   • MGI’s Stereo-seq and Olink’s high-plex proteomics extend the omics layers.
3. Biological AI and analysis (DRAGEN & Lens):
   • DRAGEN integrates SNV, indel, STR, SV and CNV calling in one framework.
   • Tempus AI’s Lens reasons over multimodal records with specialized AI agents.